This action might not be possible to undo. Are you sure you want to continue?
Railway Track Design: A Review of Current Practice
Subject Occasional Paper
This Paper reviews the current practice for the design of conventional railway track. Current track design practice has historically developed slowly and is predominantly based upon years of operating experience. Consequently many of the developed design expressions are at best semi empirical in nature. The scope of this report is confined to the design of the track structure above the level of the formation, i.e., rails, sleepers and ballast.
Series Date A to Z
Search Results Print Exit
GO BACK
BUREAU OF TRANSPORT
ECONOMICS
RAILWAY A
TRACK
DESIGN PRACTICE
REVIEW OF CURRENT N.F. Doyle Melbourne Research
BHP
Laboratories
AUSTRALIAN
GOVERNMENT
PUBLISHING
SERVICE
CANBERRA 19 8 0
@ Commonwealth of Australia
ISBN 0 642 05014 7
1980
Printed by C.I.THOMPSON, Commonwealth
Government Printer,Canberra
FOREWORD
In the design of conventional railway track the designer must select from various available procedures and criteria which influence the resultant final design. This report a is review of design methodology and examines in detail currenttrack design practices. The report contains detailed information relating to the engineering design of rails, sleepers and ballast, and where available presents alternate design formulae and assumptions. This report was prepared aas source of basic design information
included in an interactive track design model now being developed by the author. This model, which is the subject of present research, has the initial objective of producing designs for track having minimum capital any new costs. design procedures, this report
While not presenting
presents a comprehensive review of current knowledge and provides new insights into current practices. As such it should be a useful document for all those involved in railway track design
and research.
~
I
The report was prepared for the BTE Mr Neil by Doyle, Research Officer, BHP Melbourne Research Laboratories under the technical direction of Dr Ian Mair, Engineering Research Manager, BHP Melbourne Research Laboratories. The report was editedby Messrs. Neil Gentleand Chris Sayersof the BTE, who have also been responsible for BTE/MRL liaison.
CR. W. L. WYERS) Assistant Director
Planning Bureau of Canberra May 1980 Transport Economics and Technology
iii
CONTENTS Page FOREWORD SUMMARY CHAPTER 1 CHAPTER 2 INTRODUCTION THE CURRENT PRACTICE FOR DESIGNING CONVENTIONAL RAILWAY TRACKS RAIL ANALYSIS The Design Vertical Wheel Load Types of impact factor formulae currently used to calculate the design vertical wheel load The Design Lateral Wheel Load The Location of the Maximum Stresses in the Rail Section The Rail Considered as a Beam on a Continuous Linear Elastic Foundation Limitations of the beam on elastic foundation analysis The base parameter of the foundation Calculation of the Rail Bending Stress the Baseof the Rail Allowable rail bending stress at the rail base at Calculation of the Rail Bending Stress the Lower Edge of the Rail Head Calculation of the Vertical Deflection of the Rail Allowable vertical deflection of the rail Longitudinal Temperature Stresses Induced in the Rail The AREA method for calculating the longitudinal stresses The German method for calculating the longitudinal stresses at
22
27 29
iii xxii
l
3
CHAPTER 3
32 33
44
48
56
62
65
66
68
70
iv
Wheel to Rail Contact Stress Considerations The P/D ratio The calculation of the wheel/rail contact stresses and the maximum allowable shear stress in the rail head Yield and shakedown in plain strain rolling contact
73 73 75
81
88 90
91

Lateral Track Stability The critical lateral force required to shift the track CWR track buckling considerations Rail Wear Life The Universityof Illinois rail wear formula The AREA rail wear formula The Couard method of calculating rail wear Permissible limits of rail head wear CHAPTER 4 SLEEPER ANALYSIS The Effective Sleeper Support Beneath the Rail Seat
104 105
109 115
120 125 129
Area
Determination of the Maximum Rail Seat Load 132 The beam on a continuous elastic 132 foundation model The AREA method 135 139 The ORE method The applied wheel load on the rail 139 considered as being distributed between three adjoining sleepers The Contact Pressure Between the Sleeper and the Ballast The maximum allowable contact pressure The Calculation of the Maximm Sleeper Bending Moments The maximum sleeper bending moment at the rail seat The maximum sleeper bending moment at the centre of the sleeper 142 149
152 153
15 4
V
The Flexural Capacity and Requirements of Sleepers The flexural requirements of timber sleepers The calculationof the required minimum flexural capacity of prestressed concrete sleepers The flexural requirements of steel sleepers CHAPTER 5 BALLAST ANALYSIS The Determinationof the Required Ballast Depth Theoretical solutionsof the vertical pressure distribution with ballast depth Semiempirical and empirical solutions of the vertical pressure distribution with ballast depth Allowable Subgrade Bearing Pressure Safe averagebearing pressure methods British Rail foundation design method CHAPTER 6 PARAMETRIC EVALUATION OF IMPORTANT TRACK DESIGN CRITERIA Rail Bending Estimates Stress and Deflection Pressure
157 157 166
171
176
177
177 185
197 19 8 202 208 208 222 230 230 247 254 269 280
Sleeper to Ballast Estimates
Contact
Sleeper Bending Stress Estimates Effect of varying the sleeper thickness upon the maximum sleeper bending stress Ballast CHAPTER 7 Depth Requirement
CONCLUSIONS
BIBLIOGRAPHY ABBREVIATIONS
vi
P age ANNEX A THE THEORETICAL DETERMINATION OF THE TRACK SPRING RATE THE ESTIMATION OF THE COEFFICIENT SUBGRADE REACTION
OF
257
ANNEX B
266
vii
P 3.11 Criteria for Calculating the Allowable Bending Stress in the Rail at the Rail Base (Robnett et al. 1975) Allowable Fatigue Bending Stress at the Centre of the Rail Base (Eisenmann 1969b) Parameters to calculatetheAdditionalBending Stress on the Lower Edge of the Rail Head (Eisenmann 1970) Rail Joint Gap Lengths for Particular Lengths of Rail, German Conditions (Schramm 1961) Typical Values for the Track Resistance to Longitudinal Movement of the Rail,WO per metre of Rail Length (Schramm 1961) 3.16
P/D Ratios CurrentlyUsed
age
52
3.12
55
3 .l3
61
3.14
72
3.15
72 Types
for
Various
Sleeper
by RailwaysOperators75 and Recommended by Railway Organisations (Taute etal. 1971, Read 1972) 96
3.17
Comparison Between Theoretical & Experimental Buckling Results (Bartlett 1960) Typical Values of Lateral Sleepers Resistance (Shenton et al. 1973) ValuesinTable3.18PresentedasaPercentage of the Total Resistance TestSeries 1, Effect of ShoulderSizeUpon the increase of Lateral Sleeper Resistance, (Shenton et al. 1973) Test Series 1, Comparison of Timber and Concrete Sleepers Lateral Resistance for Various Ballast Conditions (Shenton et al. 1973)
3.18
98
3.19
99
3.20
100
3.21
101
24722/802
ix
Page
3.22
Test Series 2, Comparison of Timber and Concrete LateralSleeper Resistance on Granite (Shenton et al. 1973) Effect of Various Maintenance Operati,ons on the Magnitude of the Lateral Resistance of , Concrete Sleepers (Shenton et al. 1973) Summary of the Average Field Tangent Track Rail Wear Data for 57 kg/m Rails in Various U.S. Railway Systems, (Hay etal. 1973) Summary of the Average Field Tangent Track Rail Wear Data for 66 kg/m Rails in Various U.S. Railway Systems (Hay et al. 1973)
101
3.23
102
3.24
107
3.25
107
108
108
328
Rail Life 'lKCl1Factor Curvature and Use Of Curve Oilers (Danzig et al. 1976) Rail Life "KGt1 Factor (Danzig et al. 1976) Rail Life IIKR1' Factor et al. 1976)

112
3.29

 Track Gradient
Weight (Danzig
112
~
3.30
 Rail 
113
3.31
Rail Life IIKVql Factor Train Speed (Danzig et al. 1976) Rail Life "K" Factor et al. 1976)
114
3.32

 Wheel
Load (Danzig
114
3.33
Wear Factors for Calculating Rail Head Wear By the Couard Method (Prause et al. 1974)
117
X
Page 3.34 3.35
Limits of Rail Head Wear (Hay et al. 1973)
AAR Limits for Rail Head Wear (Railway Track & Structure 1970)
12 2 122
4.1
Timber Sleeper Dimensions and Spacings Currently Used by Australasian Railway Systems (Gordon 1973) Values of the Ratio 5 = grip, for Various Sleeper Types and Spacings (ORE 1969) Comparison of the Formulae Used for the Calculation of the Maximum Rail Seat Load Experimentally Determined Values of the Average Modulusof Rupture and Estimates of the Endurance Limit for Various New Hardwood Sleeper Conditions (Duckworth 1973)
128
 "
4.2
140
4.3
141
4.4
160

4.5
Experimentally Determined Modulus of Rupture Values for Various Timber Sleeper Types and Age Condition Together with an Estimate of the Endurance Limit of the New Sleepers (Reid 1973) Flexural Performance Requirement for Prestressed Monoblock Concrete Sleepers (Birman 1968) Section Properties of Proposed Range Stee 1 S leepers
of BHP
161
4.6
169
4.7
174
5.1
Safe Average Bearing Pressures of Subgrades, (Clarke 1957) Effect that the variation of Track Modulus has on the Maximum Rail Bending Stress and Deflection for Various Locomotives at 100 km/h and using 4 7 kg/m Rails
199
6.1
2 17
xi
Page
6.2
Estimates of the Maximum Rail Bending Stress and Deflection Calculated for various Rail Sections, Speeds and Axleloads, using and an Assumed Value of the Track Modulus 1 5 of Mpa
219
6.3
220 Effect of Increasing Speed and Axleload on the Maximum Rail Bending Stress and Deflection for the Range of Rail Sections considered and using an Assumed Track Modulus Value 1 5 of Mpa (per cent)
6.4
Effect of Increasing the Rail Section on the Maximum Rail Bending Stress and Deflection ' for an Assumed Value of Track Modulus of 15 Mpa Effect of Varying Rail Section and Track Moduli on the Beam on an Elastic Foundation Estimate of the Rail Seat Load, for VICRAIL XClass Locomotives and Sleeper Spacing of 6 lOmm Effect of Varying Rail Section and Track Moduli on the Beamon anElastic Foundation Estimate of the RailSeat Load, for WESTRAIL LClass Locomotives and a Sleeper Spacing of 6 1 0 mm Sleeper Length and the Design Effect of Varying Loading Conditionupon the A.R.E.A. Estimate Ballast Contact Pressure for of the Sleeper to of 610 mm a Sleeper Spacing
221
6.5
225
6.6
225
6.7
229
6.8
Effect of Varying the Sleeper Length upon the 233 Sleeper Bending Stress at the Rail Seat Calculated by the Schramm/Battelle Method, for a Sleeper Spacing of 610 mm Effect of Varying the Sleeper Length upon Sleeper Bending Stress at the Rail Seat Calculated by the A.R.E.A. Method, for a Sleeper Spacing of 610 mm the 234
6.9
xii
Page
6.10 Effect of Varying the Sleeper Length upon the Sleeper Bending Stress at the Centre for a Sleeper Spacingof 610 mm Effect of Varying the Sleeper Thickness upon the Sleeper Bending Stress at the Rail Seat Calculated by the Schramm/Battelle Method for a Sleeper Spacing of 610 m . Effect of Varyingthe Sleeper Thickness upon the Sleeper Bending Stress at the Rail Seat Calculated by the A.R.E.A. Method, for a Sleeper Spacing of 610 mm Effect of Varying the Sleeper Thickness upon the Sleeper Bending Stress at the Centre for a Sleeper Spacing of mm 610 Effect of Varying the allowable Subgrade Pressure upon the Required Ballast Depth using the Schramm Method
237
6.11
241
6.12
242
6.13
243
6.14
249
615, Effect of Varying the Sleeper and the allowable 250 Subgrade Bearing Pressure upon the Required Ballast Depth using the Schramm Method at a 600 mm Sleeper Spacing
~ . 1 Values of C O 106 kN/mm3 for Square Plates 267 1 ft X 1 ft, (300 X 300 mm) Resting on Sand (Terzaghi 1955)
B. 2
Values of CO 106 kN/mm2 for Square Plates 1 ft X 1 ft (300 X 300 mm) Resting on Precompressed Clay (Terzaghi 19551
267
xiii
FIGURES Page
1.1
Flow Chart for Conventional Ballasted Track al. 1974) Structure Design (Prause et Statistical Distribution of Measured Stress and Deflection Values, Showing the Effect of Increased Speed upon the Range of the Standard Deviation (Eisenmann 1972) The Relationship Between the ORE Impact Factor and the Train Speed €or Various Track Classes (ORE 1965, Birmann 196566) Comparison of DynamicWheelLoadson BRShowing the Relative Effects of Higher Speeds, Static Axleloads and Unsprung Masses (Railway Gazette 1970) Comparison Impact Factor Formulae of DB Locomotives and Freight Wagons, with Axleloads (kN) and Distances between Axles (m), (Birman 1966) LateralGuideForces as aFunction of theTrack24 Curve Radiusat Various Mean Speeds for DB Locomotives and Freight Wagons CBirmann 1966) Relation Between the Guide Force and the Track Curvature for Vehicles at High Speeds (Mean Values) (Eisenmann 1970) Loads on the Rail and Positions of High Rail Stresses Deformation of an Infinite Beam on an Elastic Foundation Modelling of Foundation Deflection Properties
4
3.1
9
3.2
15
3.3
18
3.4 3.5
19 23
3.6
3.7
25
3.8
28
3.9
30
3.10
35
xiv
P age
3.11
Transformation of Transversely Sleepered Track to an Assumed Longitudinally Sleepered Track for use with Zimmerman Theory (Eisenmann 1969b) Master Diagram for Noments, Pressure Intensity and Rail Depression Under a Single Wheel (Hay 1953)
38
3.12
46
Load
3.13
General Load Interaction Diagram for the Calculation of Maximun Rail Bending Moment and Deflection Superposition of Rail Stresses Caused by Vertical and Horizontal Loads to Obtain Resultant Rail Stresses Caused by a Skewed Load (ORE 1966) Ratio Between the Calculated Bending Stresses in the Rail Head and Rail Flange oK/ob (ORE 1966) Track Deflection Criteria for Durability (Lundgren etal. 1970) Half Space with a'Strip Load and the Resulting Stress Distribution with (Eisenmann 1970a)
47
3.14
57
3.15
63
,
316
67
3.17
77 Depth
3.18
Maximum Shear Stress in the Rail Head as a Function of Wheel Radius and Wheel Load (Eisenmann 1970a) Contact Surface Between Rail and Wheel (Mair 1974) Yield and Shakedown Limits in plane Strain with Combined Normal and Tangential Loading (Johnson and Jefferies 1963)
a0
3 .l9
a2
3.20
a5
xv
P
3.21
age
85
Schematic Representation of Interaction Between Applied and Allowable Contact Stress Values (Mair and Groenhout 1975) Rail Strength Requirements to Avoid Corrugation Variation of oY/ault Rolled 1978) Alloy Rail with oy for a Range of as Steels (Mair and Groenhout
3.22
87
3.23
89
3.24
Fastening Types used in Fastener Torsional Resistance Experiments (Vink1978) Torsional Resistance of Fastenings (Vink 1978) Relationship Between Sleeper Lateral Resistance and Cumulative Volume of Traffic, Showing the Influence of the Tamping Operations (Shenton and Powell 1973) Approximate Relationship Between Rail Weight 1976) and Expected Rail Life (Westrail Envelope of Rail Wear Limits for Loss of Rail Height and Side Head Wear Caused by Flange Contact: Canadian Limits (King 1976) Hypothetical Distribution of Sleeper Bearing Pressure and Corresponding Sleeper Bending Moment Diagrams for a Sleeper (Talbot 1920) PrincipalSleeperDimensions Values of the modulus of rail support Cky) for Various Axle Spacings and Assumed Track Moduli (O'Rourke 1978)
94
3.25
95
3.26
103
3.27
119
3.28
123
4.1
126
4.2 4.3
131 134
xvi
4.4
Approximate Percentage of Wheel by an Individual Sleeper
Load
Carried
4.5
Effect of Lateral Loadson Sleeper Contact Pressure and the AREA Recommended Design Assumptions (Raymond 1977) Assumed Between Bending Contact Pressure Distributions Sleeper and Ballast for Sleeper Moment Calculations
144
4.6
147
4.1
Measured Sleeper Soffit to Ballast Contact Pressure for Concrete Sleepers and an Assumed Idealised Contact Pressure Distribution (ORE 1968) Estimation of Maximum Sleeper Bending Moment at the RailSeat Determination of "Lever Arm" Vertical Stress Beneath Surf ace
h (ORE 1969) 172
14a
4.8
155
4.9
5.1
a Point Load at the l80
5.2
Application of Boussinesq's Elementary Single Vertical Concentrated Load over a Uniformly Loaded Rectangular Bearing Area Comparison Experimental of Vertical Stress Distribution with Depth and the Boussinesq Solution (ORE 1968)
183
5.3
186
Theoretical
5.4
Stress Beneatha Uniform Loadof Infinite Length (Eisenmann 1970b) Vertical Stress Transmission by Means of the 2:l Distribution
187
5.5
188
24722/803
xvii
5.6
Comparison of the Vertical Stress Distribution Under a Uniformly Loaded Circular Area Based on Boussinesq Equations and 1:l and 2:l Distributions (Departmentof Scientific and Industrial Research1961) Maximum Vertical Stress on the Subgrade (Schramm 1961) Maximum Vertical Stress u z at Depth z Below the Rail Seat According to the Empirical Comparison of Theoretical, SemiEmpirical and Empirical Stress Distributions with Ballast Depth Approximate Correlation of the Casagrande, PR and CAA Classification on the Basis of Bearing Capacity (Departmentof Scientific and Industrial Research1961) Cumulative Strain Resulting from Repeated Loading Testsof London Clay (Heath et al. 1972) Modified SN Curve (Heath et al. 1972) Derivation of the Design Chart Relationship Between Induced Stresses and Soil Strength (Heath etal. 1972) Design of Conventional Rail Track Foundations (Heath etal. 1972) Victorian Railways Main Line Diesel Electric XClass Locomotive West Australian Government Railways Main Line Diesel ElectricL Class Locomotive
193
5.7
5.8
194 Methods 196
5.9
5.10
200
5.11
204
5.12 5.13
205 206
5.14
207
6.1
209
6.2
210
xviii
6.3
Variation of the Maximum Rail Bending Stress Limit Assumed Track Moduli Values for various Locomotives at Speed, Using 47 kg/m Continuously Welded Rails Variation of the Maximum Rail Bending Stress Limit Assumed Track Moduli Values for various Locomotives at Speed, Using 53 kg/m Continuously Welded Rails Variation of the Maximum Rail Deflection Assumed Moduli for various Locomotives at Speed, using 47 kg/m Rails Variation of the Maximum Rail Deflection with Assumed Track Moduli for various Locomotives at Speed, using 5 3 kg/m Rails Estimates of the Variation of Rail Load with Sleeper Spacing for XClass Locomotives Seat VICRAIL with 213
6.4
6.5
214
6.6
215
6.7
223
6.8
Estimates of the Variation of Rail Seat Load with Sleeper Spacing for WESTRAIL LClass Locomotives Variation of the Sleeper to Ballast Contact Pressure with Sleeper Spacing Variation of the Sleeper Bending Stress of the RailSeat with Sleeper Spacing using the Battelle/Schram Method (for Variable Sleeper Lengths) Variation of the Sleeper Bending Stress at the RailSeat with Sleeper Spacing Using the A.R.E.A. Method (for Variable Sleeper Lengths) Variation of the Sleeper Bending the Centre with Sleeper Spacing (for Variable Sleeper Lengths 1 Stress at
224
6.9
227
6.10
231
6.11
235
6.12
xix
Page
6.13
Variation of the Sleeper Bending Stress at the RailSeat with Sleeper Spacing, using the Battelle/Schramm Method (for Variable Sleeper Thickness) Variation of the Sleeper Bending Stress at the RailSeat with Sleeper Spacing, using the A.R.E.A. Method (for Variable Sleeper Thickness) Variation of the Sleeper Bending Stress at the Centre with Sleeper Spacing (for Variable Sleeper Thickness) Variation of the Required Ballast Depth with Allowable Subgrade Pressure for a Range of Sleeper Spacings, using Schramms Formula, VICRAIL XClass Locomotives, and 2.440 m Long Sleepers Variation of the Required Ballast Depth with allowable Subgrade Pressure for a Range of Sleeper Spacings, using Schramm's Formula, WESTRAIL LClass Locomotives, and 2.440 m Long Sleepers Variation of the Required Ballast Depth with allowable Subgrade pressure for a Range of Sleeper Spacings using Schramm's Formula, VICRAIL XClass Locomotives and 2.640 m Long Sleepers Preliminary Track Design Interaction Diagram Lines of Equal Vertical Pressure (kPa) in the Ballastfor a Single Loaded Sleeper (Talbot 1920) The Ballast Pyramid Model
239
6.14
244
6.15
245
6.16
250
6.17
252
6.18
253
7.1
256 258
A.l
A.2
259
xx
Page
A.3
Progressive Steps in the Development Static Modelof a Conventional (Robnett et al. 1975)
of 261 a
Track
Structure
A.4
Track Depression for Different Subgrade Types 265 for an Axle Load of 190 kN (Birmann 1968)
xxi
SUMMARY
This report reviews the current practicethe for designof conventional railway track. Current track design practice has historically developed slowly and is predominantly based upon years of operating experience. Consequently many of the developed design expressions areat best semiempiricalin nature.
The scope of this report is confined to the design of the structure above the level of the formation, i.e., rails, sleepers and ballast. Conventional railway track can to some extent be regarded a as panel consisting of rails and sleepers floating on supporting ballast and subgrade layers. For any given traffic environment this panel is subjected to both dynamic and static loading conditions which are largely dependent upon the strength of track The structure and the for degree the and design quality of of track track
tra
maintenance. is based
current
practice
railway
upon satisfying several criteria for the strength of the individual
track components. These criteria, which include limitson rail stresses, sleeper stresses, the pressure between the sleeper and the ballast and the pressure on the subgrade, are extensivel reviewed in the report.
Design concepts essential to the analysis of the track components (also reviewed in detail) include impact factor determination, beam on an elastic foundation analysis, the track modulus and th rail seat' load analysis. The track component that most seriously limits the progression to higher design axle loads is the rail. Therefore, the report also includes a detailed review of rail head contact analysis and methods currently to used estimate the life of the rail.
xxii
~
At present a standard of track maintenance is impliedby the current track design procedure through the use of various factors of safety. These factors appear to be based upon the considered judgement of earlier railway researchers, and whilst the resultant track design is obviously safe no attempt is made to relate the design procedureto the expected track maintenance requirements. Consequently, for tracks that are be tomaintained to a high standard, the current track design
do not allow design procedures
I
~
benefits via the reduction of the calculated stress levels any in of the individual track components. It is also apparent that the ballast and subgrade properties need be to examined and included in the current design procedure. Allowable limits of deformation can thenbe based upon the requirenents of track qeozetry retention and therefore the standard of maintenance can be included in the design procedure.
xxiii
CHAPTER 1
 INTRODUCTION
Despite over100 years of operating experience, the design of railroad track usually depends a tolarge extent upon the engineering experience of the designer. Due to the problems of relating an applied track loading a to large number of observed track responses,
,
i
~
the
developed
design
expressions
which are
available are at best semiempirical. Therefore an understanding of the assumptions and limitations that are implicit in the use of such expressions is required. The dynamic response characteristics of the track are not sufficiently well understood to form the basisof a rational design method.At present the design of railroad track relies on relating the observed dynamic response to an equivalent factors. static response, by making use of various load
It is intended that this report willa be detailed design of conventional railway track. The objectives of this study are:
review
of
the
.
to collate the available track design expressions together with the relevant engineering rationale, and from this literature review produce a report explaining the design of conventional railway track which incorporates a listing of the relevant track design expressions to use this report as the source document for an Interactive Track Design Model which will optimise the track design parameters ona basis of cost such that sensitivity analyses can be conducted practice to identify limitations in the current design
.
.
identify areas for future design deyelopment to improve the cost efficiency of track design practice.
24722/804
1
Although the capability of the formation to resist the bearing pressures imposed by the loaded track should be analysed the scope of this study is confined to the design of the track structure above the level of the formation, i.e., rails, sleepers, and ballast.
,
The design methods currently used are generally on based estimating the stressesat various critical locations of the track structure
and comparing these stresses with suitable design criteria. It is important that these calculations be based upon sound data derived from experience with commonly used track designs, curren axle loads and operating speeds and therefore representing the actual track conditions as closely as possible.
As this reportis limited to the current practice used in the design of conventional railway tracks, the following aspects of the permanent way assembly, which have not been quantitatively related to service performance, are outside the scope of this report:
track fasteners, i.e., fish plates, rail welds, rail fasteners, sleeper bearing plates etc
. .
turnouts and crossings ballast material considerations, i.e. grading, breakdown, deformation.
However, it is recommended that available data on these aspects should be assembled for future use and completeness, particularly the ballast deformation properties which have a significant bearing on track surfacing maintenance costs.
The structureof this report generally follows the flow chart, Figure 1.1, developed at Battelle Columbus Laboratories (Prause et al. 1974). The report expands on this general design flow chart in order to develop a more comprehensive relationships between design parameters.
2
description of the inter
The conventional rail track structure is a load distributing system in which the cyclic loads associated with the passage of vehicle wheels are transmitted from the rails to the sleepers and then to the formation through a protective ballast layer. The on the formation is dependent magnitude of the stresses imposed upon the depth of the ballast layer.
The current practice for designing the railway track is based upon satisfying several criteria for the strength of individual components (Prause, Meacham 1974).Some of the important Criteria are : . allowable rail bending stress . allowable sleeper bending stress
.
.
allowable ballast pressure allowable subgrade pressure.
These criteria are generally evaluated in sequence as indicated in Figure 1.1. The main purpose of establishing these strength criteria is to reduce to a tolerable level the amount of track damage caused by particular track responses. The main track responses and the associated track damage caused by excessive loading are presented in Table 2.1.
TABLE 2.1 Track
 RELATIONSHIP
BETWEEN
TRACK FESPONSE AND TRACK Track Damage
DAMAGE(a)
Response
Rail batter and shelling Rail head contact stresses Rail shear forces and web shear Web and bolt hole failures stresses Rail fracture and fatigue Rail bending moments Sleeper loads Track displacement Ballast and Ballast Sleepers
(a) Jenkins, Stephenson, Clayton, Morlar,d and Lyon, 1974.
3
Vehicle Parameters (weight and size)
1
Assumed Roadbed Support Modulus
Rail Electrical Foundation Analysis Rail Bending Stress Availability Rail Thermal Track Stress Condition Experience, Factors Rail Yield Stress
+
,
I
Allowable Rail
m
Bending Stress

(No)
Rail Defiection
Sleeper Support Analysis Sleeper Plate Load Sleeper Bending Stress Analysis
1
Allowable Sleeper Strength Bending Stress
, 'I
(No)
Sleeper [late Load
Ballast Pressure Analysis Ballast Pressure Ballast Strength Pressure
I
I I
I
l
Subgrade Pressure Analysis
Depth
1
(yes)
I
I
Subgrade
Subgrade Pressure Allowable Pressure
I
L " 
4
/ \
Exceeded
1 Test I
Soil
Figure 1 .l Flow chart for conventional ballasted track structure design (Prause et al 1974)
4
CHAPTER 3

RAIL
ANALYSIS of a
The usual starting point for determining the suitability particular railto carry out its function of withstanding the applied vehicular loading is to calculate the design wheel load. Having calculated the design wheel loaci, the beam on elastic foundation model is then used to calculate rail the bending stresses caused by this loading. Winkler
l
(1867) and Zimmerman
this model over 100 years ago for use in the design of rails. Others who nave used rail design approaches based on this model include Talbot (191819341, Timoshenko and Langer (19'32) and Clarke (1957). The rail bending stress is usually calculated at the centre of the rail base, but the stress at the lower edge of the rail head may be critical if the vehicles impose high guiding forces during curving between the wheel flange and rail head. Temperature stresses induced in the rail are then calculated, and the total combined rail stress is compared with the allowable rail stress, which is based upon fatigue considerations. Having satisfied the bending stress criteria the rail is evaluated for excessive vertical deflection under the imposed design loading. The rail is then analysed to establish its capacity to withstand the contact stresses at the point of wheeljrail interaction. The track component that most seriously limits the progression to
(1888) developed
is the rail itself higher axle loads in railway operations (Tante, Botha1971). Due to the current trend towards higher axle loads, the selection of the ultimate strength of the rail steel has a significant influence upon the operatingoflife the rail in service.
The track panel is further analysed for buckling and lateral track stability under the expected inservice conditions. Finally an estimate of the rateof rail head wear can be empirically calculated to determine the expected of life the rail under particular operating conditions.
THE
DESIGN
VERTICAL
WHEEL
LOAD
The nominal vehicle axle load is usually measured for the stati condition, butin the design of railway track the actual stresEs in the various components of the track structure and in the rolling stock must be determined from the dynamic vertical and lateral forces imposed by the design vehicle moving at speed. The dynamic wheel loads cause increases in the rail stress valu above those of the static condition due to the following factors:
. .
. . .
lateral bending of the rail eccentric vertical loading transfer of the wheel loads due to the rolling action of the vehicles vertical impact of wheel on rail due to speed irregularities and nonuniformitiesin the track and the wheel and rail profiles.
The general method used in the determination of the design vertical wheel load is to empirically express it as a function of the static wheel load, i.e.,
P
=
!%PSI
(3.1)
where
P
p ,
=
= =
design wheel load (kN), static wheel load (kN), and dimensionless impact factor (always >l).
g4
The expression used for the calculation of the impact factor is determined empirically and is always expressed in terms of train speed. When developing expressions of the impact factor, the number of above mentioned factors considered depend upon the amount and quality of the track instrumentation used, and the assumptions used in relating the parameters. Earlier empirical expressions of the impact factor are expressed in terms of the square of the vehicle speed; thereby implying a
6
solution related to kinematic theory. Some European formulae developed for the impact factor are of this form. The main criticism of this type of empirical formulae is that they neglect any vertical track elasticity which absorbs some of the impact blow on the rail. Clarke (1957) has stated that for a given axle load, experimental results have shown that the rail flexural stresses vary with the vehicle speed V, to the power of 1 to 1.2'l) . In the following sections variocs types of expressions that have been developed for determining the impact factor are reviewed. Types of Impact Factor Formulae Design Vertical Wheel Load The main types of formulae currently used to calculate the
currently
used
to
determine
the
value
of the impact factor to enable the wheel load are discussed below.
calculation of the
design
AREA Formula: The American Railroad Engineering Association
(AREA) formula(2) quoted by Prause et al. (1974) recoxmended for determining the impact factor used in the calculation of the design wheelload is a functionof the vehicle speed and the wheel diameter, i.e. ,
"
6= 1
where V D
+ 5.21j7 ,
V
= =
vehicle speed (km/h), and wheel diameter (mm) .
Eisenmann's Formula: This method adopts a statistical approach to determine the magnitude of the impact factor. Eisenmann (1972) suggests that t ' n e rail Secding stress and deflection are normally distributed and that the mean values car, be calculated
(l) Clarke does not reference these intrack experimezts in his paper. ( 2 ) This formula was pricr ?eveloped by the fissociatioc of American Railroads (&&R).
/
from the beam on
elastic
foundation
model
which
is
discussed
in
detail later. This normal distribution is illustrated in Figure 3.1 for both rail stress and rail deflection values, (which are . obviously interrelated) Denoting the mean rail stress (or rail deflection) X the by Corresponding standard deviation of this mean value S, can be expressed by
where
c: = a Q = a
factor dependent upon the track condition, and speed factor.
The value of 6 is determined by the quality of the track, and the following
6
values
have
been
suggested
for
use:
=
0.1, for track in very good condition 0.2, for track in good condition 0.3, for track in poor condition.
B 6
= =
The
valueof
Q
is determined by the speed of the vehicle, V
(km/h), and the following values
rl
Q
have
been
suggested
for use:
=
=
1, for vehicle speeds up to60 km/h
l + v6o 140' €or 200 km/h.
vehicle
speeds
in
the
range 6 0 to of
The product &.Q is referred to as the coefficient of variation. The corresponding maximum rail stress (or rail deflection)is given by
X
=
;;+st,
(3.4)
where t depends upon the chosen upper confidence limits definin the probability that the maximum rail stress (or rail deflection)
8
Speed
Note:
M a x i m u m Value = M e a n Value X (1 + where t = Student ' t ' statistic F = standard deviation of the m e a n
t s i
Figure 3 . 1 Statistical distribution of measured r a i l stress and deflection values,showing the effect o f increased speed upon the range of the standard deviation, (Eisenmann 1972)
24722/805
9
will not be exceeded. The maximum rail stress (or rail def ection), X can be defined by the simple relationship
X
where
X
=
bx l
3. S)
 =
=
mean rail stress (or rail deflection) from beam on elastic foundation model, and impact factor relating to the track condition and train speed.
g5
Combining Equations3.3 and 3.4 and equating to Equation 3.5 the expression for the impact factor reduces to
where values of t for the chosen upper confidence limits (UCL) can be obtained from the following: t t t
t
=
0, UCL
1, UCL 2, UCL
=
50 per cent
=
= =
=
= =
84.1 per cent 97.7 per
3, UCL
cent 99.9 per cent.
ORE Formula: By far the most comprehensive method to determine the impact factor is that developed by the Office of Research Experiments of the International Union of Railways (ORE) and is entirely based upon measured track results (ORE 1965). The impact factor is defined in terms of three, dimensionless speed coefficientsa', 6' and y', i.e.,
where a' and 6' relate to themean value of the impact factor and
y' to
the standard deviation of the impact factor.
1.0
The coefficienta' is dependentupon:
. .
.
the level of the track the suspension of the vehicle the vehicle speed.
The correlation between the value 2 'of and the influence of the level of the track is difficult to estimate due to in errors measurement. In a perfectly levelled track a' is virtual ly zero. In tangent track with levelling defects and very fast traffic 35 per cent. Whereas in curved track was found to approach values of a' did not exceed 18 per cent. For the most unfavourable case increases with the cube of the speed and for the locomotives examined it was empirically expressed as
a'
= =
0.04
3 v C m )
I
(3.8)
where
V
vehicle speed (km/h).
The numerical coefficient (in this case 0.041 is dependent mainly on the resilience of the vehicle suspension. The coefficient a ' is dependent upon:
. .
.
the vehicle speed superelevation deficiency of the track the centre of gravity of the vehicle.
The coefficient 5' is the contribution resulting from the wheel load shift in curvesl and may be defined by either:
.
the French (SNCF) formula
8'
=
2d.h 7
1
(3.9)
.
the German (DB) formula
6'
=
= = = = = =
V2(2h+c) 2c.h 127 Rg
g
2
'
(3.10)
where g h d c R V
gauge width (m), height of the centre of gravity of the vehicle (m), superelevation deficiency (m), superelevation (m), radius of curve (m), and vehicle speed (km/h).
is The two formulae are approximately equivalent. The DBformula strictly accurate, whereas the SNCFformula is approximate to within about1 per cent.
Birmann (19651966) states that for the SNCF operating conditions with a superelevation error of 150 mm, the value of 6' from Equation 3.10 ranges from 0.13 to 0.17.
It was observed, under otherwise equal conditions, that the measured coefficient a' in tangent track is in almost all cases larger than the value (cY,'+~') for curved track measured and 6' calculated). I t would therefore seem more appropriate to only consider valuesof a' as the mean value of the impacts factor for
use in design. The coefficienty' is dependentupon:
. .
. . .
the the the the the
vehicle speed age of the track possibility of hanging sleepers vehicle design maintenance conditions of the locomotive power units.
The measured coefficient y' increases with speed, and, a asfirst approximation the following formula can be used if experimental data arenot available. 12
where
V
=
vehicle speed (km/h) .
If the effects of other variables are to be incorporated the above formula can be generalised as
y 1
=
=
yo.a0.b0
’
(3.12)
where a
yo
value determined by Equation 3.11, a locomotive factor relating to the maintenance condition (including the effects of locomotive aye), and a track maintenance factor relating to the standard of the track.
a ,
= =
bo
yor. a
of the coefficients The ORE (1965) have obtained typical values and bo for various track and locomotive conditions.
Implied in these values are the standards of track and locomotive maintenance that allow particular operating speeds be to safely of these coefficients maintained. The ORE emphasise the values were determined entirely from the average observed relative increase in the standard deviation of the mean rail force level. The following values of y , , aol and bo were recommended:
.
for normal track with a maximum permissible speed of up to 140 (km/h) :
YO
=
O.lll
(1)) (from observations
locomotive maintenance factor, . a = 2.0 track maintenance factors, bo = 1.3
.
for special track with an authorised speed of 200 km/h, assuming new vehicles:
(l) This compares with0.15 obtained from Equation3.11.
13
yo
=
0.24,
(from Equation3.11)
locomotive maintenance factor, 0 a = 1.5 track maintenance factor, bo = 1.2. The firstof the above mentioned values are those most
,
relevant
to Australian conditions. A maximum value of the track maintenance factor, bo was also determined for a track with relatively poor
ballast compaction beneath the sleepers. Under load this track was observed to have 3 mm voids between the sleeper and the ballast, and this condition coupled with high speed traffic gave a maximum value of bo equal 1.7. to
In summary, the ORE have observed that the maximum value of the impact factor occurs in tangent track. Consequently, the formula for the impact factor6, Equation 3.7, reduces to
# =
1 + c l ' + y'.
(3.13)
Using Equations3.8 and 3.12 and the ORE recommended values of bo and yo from test data for normal tracks with permissible speeds of up to 140 km/h, the maximum value of the impact factor (1) can be estimated by
# =
where
V
1.29 + 0.04
3 V (100)
(3.14)
~
=
train speed (km/h) .
The relationship between the maximum value of the impact factor and the train speed has been plotted for various standards of track and is presented in Figure 3.2.
(1) This compares with V 6 = 1.26 + 0.08 using Equations 3.8, 3.11 data for ao, bo and yo.
(m)
and 3.12 in the absenceof test
14
1.8
r
1.75 $
1.7 
i
1.6 a
4

5
m
m
1.5
LL
t
 1.4
1.2
1.3 
I
I
1 .l
1
1 .o
I
60
I
I
i
1
l
0
20
40 80
100 120 Train Speed V (km/h:l
140
160
180
200
Maintenance quality of Legend: Maintenance quality of Track for m a x . speed of 140 kmi'h Without levelling defects and depressions
the
locomotive
for m a x . speed of
200 kmih
" l 1
+
W i t h usual levelling defects
without depressions With usual levelling defects and depressions
"
Figure 3.2 T h e relationship between the ORE impact factor and the train speed for various track classes (ORE 1965)(Birmann 196566)
A l l these formulae suffer from the basic limitation in that they
have no adequate theoretical backing, and therefore extrapolati is not reliable while little published data exist on the statistical variation of loading under Australian operating conditions. British Railways Formula: A simple model for a discrete irreyu
larity such as dipped rail joint has been developed which illustrates the combined effect of vehicle speed, unsprung mass and track irregularities (Jenkins et al. 1974, Railway Gazette 1970, Koffmann 1972) . The model was developed for current BR main line track consisting of 54 kg/m continuously weldedrails and with concrete sleepers spaced at 760 mm. The resultant dynamic wheel load rail joint was determined from due to a wheel striking a
d
where p ,
=
static wheel load (kN), unsprung weight at one wheel (kN), track stiffness at the joints (kN/mm), 2 gravitational constant (m/S 1 ,
P , D j g
= = =
(“1 + “ 2 ) = total rail joint dip angle (radians), and V = vehicle speed (km/h).
Therefore
the
dynamic
8.784
factor can be defined as 0.5 (a1 + a 2 ) V
(3.16)
6
=
l +
For BR conditions and with a track consisting of continuously welded 54 kg/m rails on concrete sleepers spaced at 760 mm, a value of D equal to 8 8 kN/mm is considered sufficiently accurate j for use in Equation 3.15 (Koffman, 1972). Also for these conditions a rail joint dip of 10 mm and a corresponding value of a1 + a 2 0.015 radians can be used as representative values of the welded joint.
16
The resultant dynamic wheel load to due locomotives striking a dipped rail joint at speed, and estimated by Equation 3.15, has been evaluated for a range of currently used British Rail locomotives, and is presented in Figure 3.3. It can be Seen that the dynamic wheel load caused by poorly maintained joints be can as high as two to three times the static wheel load depending upon the speed of the locomotive. The four main types of impact factor formulae, so far are compared inFigure 3.4. discussed,
Although the formulae are not
specifically interrelated a few general observations about the predicted magnitude of the impact factor can be made. The of impact factor for very envelope definedby Eisenmann's curves good and good track conditions (at UCL 'of 99.9 per cent), contains both the AREA and ORE impact factor curves which have been derived for probable average track conditions. Also the impact factor curve corresponding to Eisenmann's poor track condition is approximately of the same order of magnitude as the British Rail formula for locomotives striking poorly maintained rail joints. Other developed expressions that have been to used determine the magnitude of the impact factor are presented in Table 3.1. The Indian Formula (Agarwal 1974) attempts to relate the track condition to the impact factor by making use of the measured values of the track modulus. The German Formula (Schramm 1961) is a typical expression developed solell7 on kinematic considerations. The South African Formula (Lombhrd 1974) of is the same form as the AREA Formula (Prause et al. 1974), Equation 3.2, but calculated for narrow gauge track structure. The Clarke Formula (Clarke 1957) is simply the algebraiccombination of the AREA Formula and the Indian Formula and as such not is modelled on any experimental results. This emphasises the problem of using highly empirical formulae without adequate knowledge of the track conditions and assumptions used in their derivation.
24722/806
Speed (m.p.h.)
I
50
l 175 75
l 150 100
1 125
I
1
I 200
i
225
Speed (km/h) Static Axle Load Unsprung (tonnes) (tonnes) 23 20.45 25 22.5 17.6 17.5 16.5 13.5 4.25 4.22
Mass
Legend:
No.
Vehicle KESTREL (DieselElec)Loco, COCO
Class 86 Elec. Loco, 8080
Tonne 100 Wagon 45Tank Tonne Class 55 Deltic Diesel Elec. Loco, COCO
Tank Wagon
2.8
2.7 3.3 1.92
Freightliner Wagon
Unit Special Elec. Multiple 6 Axle Freightliner Wagon
2.0
1.9 1.5
9
Advanced Passenger Train
12
Figure 3.3 Comparison of dynamic wheel loads on B R showing the relative effectsof higher speeds, static axleloads and unsprung masses (Railway Gazette 1970)
18
L
c
0
U
m
Legend:
No.
1
2
Formula Area Area Eisenmann Eisenmann Eisenmann Ore
Remarks
Wheel diameter = 900 m m Wheel diameter = 1000 m m
Very good track UCL = 99.9% Good track U C L
3 4
5 6 7
= 99.9% = 1.5, b , = 1.2
65 t
Poor track U C L = 99.9%
p'
assumed = 0.20 a ,
B.R.
Class 55 Deltic Diesel Elec. PS = 86.2kN Pu
Kestrel Diesel Elec. PS = 112.7 k N Pu
8
B.R.
= 2.12 t
Figure 3 . 4 Comparison of impact factor formulae
TABLE 3.1
 SUMMARY OF
EXPRESSIONS
THAT
HAVE
, .
BEEN
USED TO DETERMINE
THE VALUE OF THE IMPACT
1.
FACTOR(^^
Indian Formula (Agarwal 1974):
d = 1 +
2.
v
58.14(k)OS5
German Formula (Schramm 1961): (a) For speeds up to 100 km/h 2 6 = 1 + v 4 3 X 10 (b) For speeds above 100 km/h
f5 = 1 +
3.
107 South African Formula (Lombard 1974), (Narrow gauge track): 4 = 1 + 4.92 V 5
 1.5v 3 105
4 5 v '
"
4.
5.
(Clarke 1957) : 19.65V d = 1 + D(k) " * ' WMATA Formula (b) (Prause et al. 1974): d = (1 + 3.86 X 105 V 2 )0.67
"
Clarke
Formula
= vehicle speed (km/h), k = track modulus (MPa) , and D = wheel diameter (mm). (b) WMATA is the abbreviation of the Washington Metropolitan Transit Authority.
(a) Where V
A comparison of the vehicle and track parameters that have bee included in formulae developed for the impact factor are presented in Table 3.2. It is apparent that the early formulae developed to determine the impact factor are rather simplistic in that they relate only to vehicle parameters (e.g. train speed and wheel diameter). Although the track maintenance conditions is somewhat implicitin the early empirical formulae, it is only recently that attempts have been made explicitly to determine i significance tothe value of the impact factor.
20
I
TRAIN SPEED WHEEL STATIC
LOAE
~
DIAMETER TEEET, WHEEL
" "
WSPRUNG
>lAsS
CENTRE OF GRAVITY OF VEHICLE LOCOMOTIVE MAINTENANCE CONDITION TRACK MODULUS
____TRACK STIFFNESS AT RAIL JOINT
~
Il
1
I
TRACK ANGLE
JOINT
DIP
L
" " " " j _
CAXT DEFICIENCY IN CURVES
I CVRVE RADIUS
I
TRACK VAINTENANCI
THE DESIGN LATERAL The
WHEELLOAD
magnitudeof the lateral guiding force imposed on the rail
head can be considered to be dependent upon the following:
. . .
.
the curve radius of the track the vehicle speed the length of the vehicle wheelbase and its bogie configuration the tracking motion of vehicles in the train consist.
Few intrack test programmes have been carried out to determine the magnitudeof the lateral forces caused by the wheel flanges of vehicles contacting the rail head when negotiating curves. Of the published literature the results of Birmann (19661, the ORE (1965, 1970), and Olson and Johnsson (1960) appear to be of the mostuse.
Birmann (1966) has carried out a series of experiments to determine the magnitude of the lateral guiding force by caused the wheel flanges of vehicles (in particular locomotives) contacting the rail head when negotiating curves. Results of guide force measurements for various locomotive and wagon bogie configurati shown in Figure 3.5 are plotted against curve radius in Figure 3.6. It can be seen that the guide force, H(kN) is to a degree more dependent upon the curve radius than the vehicle speed. for radii greatee than 800 m and at the same speed, the magnitude of forces exerted by a given vehicle are similar to those exerted on tangent track due to the tracking motion of the train consi However with radii less than 800 m the lateral forces increase significantly with decreasing radii. The recorded maximum values of the guiding force were found to be of the order of to 30 60 per cent higher than the mean of the recorded values. The relationship of the mean lateral guiding force exerted by various to Eisenmann (1970) and is shownin Figure 3.7.
vehicle types to the curve radius of the track has been presen
22
208
208 208 208
188 188 188
188
188
E 03
B R 50
4.7
15.90
4i
D J
p & 8 J 2J j
12.10
Locomotives
E 10
v
100
Ommi 51
11.50

13.20
Ootz 50
Freight Wagons
Ssym 46
Figure 3.5 DB locomotives and freight wagons, with axle loads (kN) and distances between axles (m) (Birmann 1966)
23
60
Locomotives 7
= 60 km/h
km/h
v = 20
I
I
I
c70 0 500 1000 Curve Radius(m) BR50 Leading Axle
0
500 1000 Curve Radius (m) E50 Leading Axle
V = 120 km/h
V = 80 kmlh
V = 80 km/h
V = 60 km/h
V = 20 km/h 20
0
U
0
500 1000 Curve Radius (m) E 1 0 Leading Axle ( m )
V = 120 km/h V = 80 km/h
0
500 1000 Curve Radius(m) BR50 Sixth Axle
2
Y
60
V = 60 km/h
p
0 LL
40
20
20 0
0
c7
0
500 1000 Curve Radius (m) V200 Leading Axle
V = 100 km/h V = 80 km/h ,V= 60 kmlh
0
0
500 1000 Curve Radius(m) V100 Leading Axle
30
20
Freight Wagons
v
0
1000 500 Curve Radius (m) Ssvm 46 Leading Axle
0
5
W
1
1000 500 Curve Radius(m) Ommi 51 Leading Axle
V
= 80 km/h
2
LL
Pc7
W
10 0
500 1000 Curve Radius( m ) Ootz 50 Leading Axle
Figure 3 . 6 Lateral guide forces as a function o f the track curve radius at various mean speeds for DB locomotives and freight w a g o n s (Birmann 1966)
24
Curve Radius (m)
t
108
1
6
5
4
3
2
1.5
Degree of Curve
Figure 3 . 7 Relation between the guide force and the track curvature for vehicles at high speeds (mean values) (Eisenmann 1970)
25
247221807
The ORE (1965, 1970) used Birmann's results and carried out an
expanded test programme for speeds up to 200. km/h. The expression suggested by the ORE for determining the magnitude of the lateral force H(kN), caused by the wheel flanges of locomotives contactin the rail head when negotiating curves is dependent only upon th radius of the curve R(m). It is based upon the observed maximum envelope of experimental results for all vehicles and is expresse as
H
=
7400 3 5 + 7

A series
of field tests were also carried out by the Swedish Railways (Olson eta1 1960) to determine the magnitude of locomotive guiding forces. Measurements of the magnitude of the lateral force caused by the wheel flanges of the locomotive contacting the rail head were obtained from a small of number tests and only for a curve radius of m. 600 Since the curve radius was kept constant the empirical expression of the magnit of the mean lateral force Hmean (kN) is dependent only upon t speed of the locomotive V(km/h), i.e.,
V = 1 7 + 27.6
It was specifically
that the magnitude of the
(3.18) mean
noted
lateral
force increases in a similar way for both the light and heavy axle load locomotives used in the tests. It is usually assumed that the magnitude of the lateral force would increase more rapidly in the case of heavy axle load locomotives. The lateral forces caused by the wheel flanges of locomotives
contacting the rail head when negotiating curves have also been investigated by British Rail (Koffmann 1972). Results of the values of the magnitude of lateral guiding forces imposed upon
26
the
rail
head
by
BoBo COCO and type
locomotives
for
various
locomotive speeds and wheel loads and ranges of curve radii are presented in Table 3.3 (1). TABLE 3.3
 GUIDING
~~ ~
FORCES IMPOSED ON TRACK BY TWO TYPES OF TRAVELLING AT SPEED ON VARIOUS CURVE RADII
LOCOMOTIVES
(Koffman 1972)
~~~ ~~
Wheel Locomotive Type Load P (kN)
Guiding Force H (kN) Tangent Maximum Curve Radius (m) 650 350 180 Track Permissible Values(km/h) Speed 40 60 60 80 100 110 110 120 52
48
115
BoB0 COCO
82 78 96 96
61 52 74 62
83 77 87 66 83 75 73 87 98 90 93 88 98 104 98 101 101 113 113 77
99
100
135
BoBO COCO
64 55
100
70 106 80 107 65 113 69
86
110 100
Having determined the design vertical wheel load and the design lateral wheel load for a given vehicle, the next step in the , design procedure is to determine the resultant maximum rail stresses caused by this imposed loading. THE LOCATIONOF THE MAXIMUM STRESSES IN THE RAIL SECTION
l
The location of the maximum rail stresses are shown in Figure 3.8 (Eisenmann 1969a). The evaluationof each must satisfy the corresponding design criteria for the allowable rail stress. The maximum shear stress occurring in the web is not usually considered
(1) Clark (1973) has defined these locomotive axle and bogie configurations and presents a summary of Australian locomotives for comparison with Figure 3.5.
27
Legend: A, B, C are areas of critical stress. P and H are applied vertical and horizontal wheel loads respectively.
Loads
Figure 3.8 on the rail and positions of high rail stresses
28
in the current rail design methods, since is it unlikely that the rail will be subjected to a shear failure (unless there is an occurrence of a rail fatigue defect, e.g. transverse defect). The bending stress which occurs at the centre of the rail base (point A) is independents of the magnitude of the guide force and the eccentricity of the point of attack of the wheel. The design criteria for this bending stress is established to prevent the occurrence of cracksin the rail base. The bending stress at the lower edge of the railhead (point B) is important in the evaluation of plastic deformation of the rail head in the horizontal direction. High values of rail shearing stress are generated near the contact pointbetween the rail and the wheel (point C) as a result of constant repetitive load introduction. When the fatigue strength is exceeded, fracture of the rail head occurs. This is commonly termed shelling of the rail head and problems of this nature will be discussed in detail later. In order to facilitate the calculation of the rail bending stress at the centreof the rail base and also the amount of vertical rail deflection under load, it is useful at this stage to introduce the concept of the rail considered a,sa beam on a continuous linear elastic foundation. THE RAIL CONSIDERED AS A BEAM ON A COPI'TINUOUS LINEAR FOUNDATION ELASTIC
~
~
~
,
i
~ ~
The concept of a foundation modulus to represent rail the support was first introduced by Winkler (18671, when he analysed the rail as an infinite beam supported on a continuous linear elastic foundation. The differential equation for the bendirg theory of an elastic beam from Figure 3.9(a) is
29
U n d eB fe oa rm e d
Axis
Distributed load q(x)
Deformed B e a m
\
Surface contact pressure p(x)
Figure 3 . 9 ( a ) Equilibrium positiono f a deformed b e a m subjected to load q(X)
Distributed load q(x)
X
Elastic foundation represented by spring base
Figure 3 . 9 (b) Representation o f a continuously supported infinite beam on an elastic foundation subjected to load q(X)
Deformation of an infinite b e a m on a n elastic foundation
30
where y(x) = vertical deflection at X, q(x) = distributed vertical load, E1 = flexural rigidity of the rail, p(x) = continuous contact pressure between the sleeper and ballast, p(x) k Hence the
= =
ky(x) , and modulus of the foundation. equation becomes (3.20)
Winkler
This
equation
may
be
represented as the
responseof an infinite
beam attached to a spring base, subjected to a load q(x) , Figure 3.9(b). The general solution of the Winkler equation has been developed in detail by Hetenyi (1946). Since the rail is subjected to wheel loads, which are loads, the relevant solution to Winkler's equation must be of the design wheel load, P, instead of restated in terms q(x). The solution of the rail deflection rail shear rail bending moment at any position X, (X positive), from load pointare: concentrated load force and the
.
rail deflection (3.21)
.
rail shear force (3.22)
31
.
rail bending moment
(3.23)
Here B includes the flexural rigidity of the beam as well as t elasticity of the supporting medium, and is an important factor influencing the shape of the elastic beam. For this reason the factor @ is called the characteristic of the system, and, since / B is frequently referred its dimension is (length)', the term ' to as the characteristic length. Consequently, the product @X will be a dimensionless number with
(3.24)
B
where k E
I
=
= = =
(m)
I
0.5
track modulus ("Pal, Young's modulus of the rail steel(MPa), and 4 rail moment of inertia (mm 1 .
The Winkler equation was originally developed for longitudinall sleepered track, and has since been applied to transversely sleepered track, thereby raising questions concerning the validit of the assumption of continuous rail support. But although there have been many methods developed to analyse track on the basis of discrete elastic supports, notably by Schwedler (1882), Zimmerman
(1888) and Engesser (18881, according to Ken (1976) the results
obtained are not significantly different from those using Winkler model. Hence considering the rail a as beam on a
the
continuous linear elastic foundation is generally regarded as the most acceptable method for the analysis of rail stresses and deflections. Limitations of the beam on elastic foundation analysis
The following are the main limitations of the on beam an elastic foundation analysis as applied to railway track conditions:
32
.
the model neglects any continuity or coupling of the ballast and subgrade layers that make up the track foundation. The magnitude of the coupling effect depends upon the sleeper spacing, the sleeper size, the ballast depth, and the subgrade properties (Eisenmann, 1969a)
.
there is no adequate modelling of the stressstrain behaviour of the ballast and the subgrade. Thus the model is of limited value in considering the behaviour of the substructure beneath the rail (Robnett, Thompson, Hay et al. 1975) the simple Winkler model does not include several additional factors which are known to affect the stresses and deflections in railroad track. These include longitudinal loads lYom thermal stresses, a restoring moment proportional to the rotation of the rail and sleeper, the eccentricity of the vertical load on the rail head, and track dynamic effects, such as inertial and damping forces (Prause et al. 1974).
.
But despite these deficiencies the Winkler model has proven quite useful for design purposes because of its simplicity, its ease of use and its degree of accuracy when ccmpared with measured results. The base parameter of the
(1) foundation
Including the foundation modulus as used in the on beam an elastic foundation analysis (Equation 3.26), there are currently three alternative ways to define the base parameter of the foundation from field measurements of railway track (Bhatia,
i
Romualdi
and
Theirs 1968) .
(l) The term base parameter refers to the assumptions used in the analysis to define the elastic foundation support beneath the beam. The general solution to Equation 3.19 for beam deformation must be redefined for each set of ass.mptions to ensure consistent dimensions.
247221808
33
Spring ConstantMethod: parameter is analagous
This method used to define the base to the spring constant used in
vibration
theory. (Figures 3.10(a) and (b)). For the track structure loaded by a known static axle load, (i.e. twice the known static wheel load PS on each rail) the total rail
Y, is the sum of the rail deflections occurring support deflection at each sleeper location to the left and of right the position of the known static axle load.
The total rail support deflection can therefore be expressed
a
m=.Y =
m=O
c
Yn m,
+
(3.25)
where Ynkm
=
rail deflection (nun) at the r n t h sleeper location away from the position of the applied axle at load sleeper n.
The foundation spring constant D (kN/mm) for the rail support derived from this mechanical model of the railway track is D where =
p ,
Y '
(3.26)
Y
= =
known static wheel load on each rail (kN), and total measured rail support deflection (mm).
Kerr (1976) defines the spring constant in terms of the track subjected to a loading of 2Ps (i.e. the axle load) as twice the spring constant determined from only one rail being subjected a load of P S I i.e.,
Dtrack
 2Drail =
20
(3.27)
Talbot (19181935) observed that the static track deflection caused total by a single of point load on the and rail nine head spread in over a number between seven
34
sleepers
track.
m,
Figure 3 . 1O( a) Conventional railway track structure
Spring
Figure 3 . 1O ( b) Mechanical model for track
S
nl
"
S
n2
n+l
I,
S
l
;
S
"S
n+2 =L
Figure 3 . 1O ( c) Continuous deflection curve
Figure 3 . 1O ( d) Discrete deflectioncurve
Legend:
S = Sleeper Spacing n k m = Sleeper Location away from Loaded Sleeper p, = K n o w n Static Wheel Load
Modelling of foundation deflection properties
35
Kurzwell (1972) states that for average track structures the
deflection due to a point loadon a track is negligible beyond about 3.000 m of the point of load application. Adopting a typical sleeper spacing, this suggests that about nine sleepers resist the applied loading. Hetenyi (1946) also reports the results of experiments conducted by Wasiutynski (1937) who noted that the value of D determined by means of Equation 3.36 for the track assembly is about half the value obtained when a single separate sleeper is loaded and is to due foundation interaction between adjacent sleepers, thus only
=
Dtrack
Dsleeper 2
(3.28)
There is a marked decrease in the track stiffness, D, in the vicinity of a fishplated rail joint (Meacham,and Ahlbeck 1969). Values of the average track stiffness at the rail joint D range j 0.77D for a from 0.25D, for a joint in a very bad condition, to joint in excellent condition. It can clearly be seen that the rail joint is the weak link in the track structure.
Track Modulus Method: This method is analagous to Young's Modulus used in determining the strength of materials. The continuous support deflections, shown in Figure 3.10(c), can be approximated by a series of stepped deflections, considered to be constant over the length of the sleeper spacing. This stepped deflection curve is presented in Figure 3.10(d). Using this assumption the sleeper spacing can be introduced into the expr to determine the rail support parameter. The track modulus k (MPa) is defined as the force/ unit of delfection/unit of track l e n g t h ' ' ) . For the case of a single rail it can be expressed as
( 1 ) The track modulus k is sometimes the railway literature.
denoted by the symbolU in
36
k where
PS
=
(3.29
Y
S
103
=
=
Y
S
known static wheelload on each rail (kN), total measured rail support deflection (mm), (by Equation 3.25) , and sleeper spacing (mm). the
=
This is the method used by AREA the and Clarke to determine rail support base parameter for use in the bean an elastic on foundation analysis.
Coefficient of Subgrade Reaction Method: This method is based upon the original Zimmerman theory which was developed a for longitudinally sleepered track compressible foundation. Since a transformation of this track sleepered track is required in if the assumption is made that (Eisenmann, 1969b) .
For transversely
considered to be resting on a track is now transversely sleepered, type to an equivalent longitudinally the analysis. This can be achieved the effective rail support area for both types of track.
provided by the sleeper remains constant
sleepered track the effective sleeper support
2
area beneath one rail seat, As (m), can be assumed to be
~
where B
R
= =
g
=
sleeper breadth (mm), sleeper length (mm), and distance between the centre line of the rail seats (mm).
This is discussed in detail later. Referring to Figure 3.11 the breadth B' (mm) of an equivalent longitudinal sleeper supporting one rail be cancalculated (assuming constant rail support area) as
37
ned Equivalent Longitudinal Sleeper
L . I u " . I I
1 . 1
Track Centre Line
Legend:
S = Sleeper Spacing
Q
= Sleeper Length
B = Sleeper Breadth
g = Distance Between Rail Sleepers Equivalent Longitudinal Sleeper B' = Breadth of = B ( Qg)/s
Note: Effective rail support area = B ( Q g)
Figure 3 . 11 Transformation of transversely sleepered track to an assumed longitudinally sleepered track for use with Z i m m e r m a n n theory (Eisenmann 1969 b)
38
where S
=
sleeper spacing (mm).
3
The coefficient of subgrade reaction C (kN/mm ) for one rail is then defined as (3.30) where PS
Y
S
= = = =
known static wheel load on each rail (kN), total measured rail support deflection (mm), (by Equation 3.25) , sleeper spacing (mm), and assumed width of the effective rail support in the transverse direction (mm).
B'
Under German Railway(DB) track conditions the sleepers are spaced 630 mm apart and the effective rail support area per rail, 5 2 is usually assumed to be 2.6 X 10 mm . Therefore the equivalent breadth of an equivalent can be expressed as longitudinal sleeper supporting rail one
It is important to realise thatthe above methodsto determine the base parameter of rail support are interrelated since
D
=
Sk
=
SB'C
(3.31) calculation
and that the only field measurement carried out is the of the value of Y for a known static axle load.
Values of the track modulus k have been proposed by Hay (1953) and Ahlf (1975) for various track structures (Tables 3.4 and 3.5). Westrail (1975) have published neasured values of track modulus for current mainline track structures (Table 3.5). The results of German measurementsof the track stiffness D have been 39
published by Birmann (19651966) (Table 3.7).
Birmann also
reports the results of Luber (1961) for the stiffnesses of individual track components and these are reproduced in Table 3.8. Typical values of the coefficient of subgrade reactions have been proposed by Eisenmann (1969b) for German Railways (DB) track conditions (Table3.9). The main factors influencing the (Lundgren, Martin and Hay, 1970): value of the track k modulus are
. .
the sleeper spacing, dimensions and quality the quality, depth and degree of compaction of the ballast layer which defines the solidarity of the track construction the subgrade quality and the degree of its compaction which determines the strength of the foundation
.
.
the rail size affects the load distribution of adjacent sleepers in the track panel.
As a first attempt to determine the significance of some of the above factors which influence the track spring rate (and therefore 1 from Equation 3.31 the track modulus)a theoretically derived ballast' pyramid model has been suggested by Prause et al. (1974). This theoretical method of determining the track spring rate is presented in Annex A.
The track modulus as defined is strictly a static parameter and is not intended to include any dynamic effects such as frequency dependent damping or the mass of the rail support system (Eisenmann 196933). Having estimated a value of the track modulus the maximum rail bending stress at the rail base, Figure 3.8 point A, can be calculated by using the beam on an analysis. elastic foundation
40
N
TABLE 3.4
m
,
 VALUES
X X X
OF TRACK MODULUS (HAY 1953)
0
\ o
I

Ballast Condition Modulus Track and Track Sleeper Size Rail (kg/m) 42 178 42 178 42 152 42 152 1.78 42
b k .
(mm) 229 229 203 203 229 229
X X X
(MPa)
2590 2590 2440 2440 2440 2440 2590 2590 2440
X X
X
X
42
12 fi 4 R .15 .(a) 5I ; 11 . E .(a)
178
X
X
tl
1.78 X 223 l78 X 229 178 X 229
G.ki.0.
X X X
560 mm centres
fastenings(b) 229
X
55 R.E (a) l78
X
2440
560 mm centres
G.E.0.
fasteninys(b)
150 mm fine cinder ballast in poor condition 3.7 clay subgrade 150 mm cinder ballast, fair condition, loam 5.2 clay subgrade 150 mm limestone on loam clay subgrade, good 6.7 before tamping 7.5 300 mm limestone on loam clay, subgrade after tamping 7.3 300 mm limestoneon loam clay subgra,de, good before tamping 300 mm limestone on loam clay subgrade, good 7.3 before tamping 0.3 600 mm crushed limestone on loam and clay 20.0 20.7 600 mm gravel ballast; plus 200 nun heavy limestone on wellcompacted subgrade Flint gravel ballast on wide stable roadbed 17.3, 17.9, 24.8 A V 20.0 25.5, 38.0, Limestone ballast on wide stable roadbed 42.8 AV 35.4
on loam

(2)
R.E.
(b)
The G.E.O. fastening used by D.13.
indicates an AREA rai.1 specification. is a German manufactured fastener and is very similar the kto fastener
Source: From the first and sixkh progress reports of special committee on stresses in Railroad Track (Talbot 1918; 1934).
TABLE 3.5
 TRACK
~
MODULUS
VALUES FOR FIVE
DIFFERENT
TYPES OF
TRACK, (AHLP 1975)
~
Sleeper Condition Poor
Ballast Ballast Condition Depth (mm) Relatively 150 unsound Poorly drained, material, fouled with mud 150
Subgrade Condition
k (MPa)
6.9
,
soft
Fair Good 300 Good
Fair soundness, Average, some reasonably free of mud drainage
13.8 20.7 27.6 34.5
150 Sound, crushed stone, Average, some drainage mud offree
Sound, crushed stone, Average, some offree drainage mud well Clean, sound, crushed stone Good, compact,
450 Good drained
TABLE 3.6
 MEASURED
VALUES OF TRACK MODULUS, STANDARD GAUGE WINLINE TRACK WESTRAIL (at 1976) of Test Sleeper Ballast Season Crs. (mm) (MPa) Depth (mm)
k
Location Date
6 1972 Aug Yard Avon Southern Cross Koolyanobbing 30.8.72 255m.16~. Southern Cross Koolyanobbiny 31.8.72 278m.00~. Bonnie Vale S tewart 372m.67~. Forrestfield 5m. Koolyanobbing West Kalgoorlie 370m. 75c. Forrestfield Cockburn Sound (Walliabup 12.9.73
10
Winter 250 (moist)
13.2
610 15.2 Winter 250 (dry) 610 13.8 Winter 250 (dry) 4.9.72 Winter 230 610 (dry) 627 617 21.3 Spring 12.7 Winter 14.8
0.72 50c.
12.9.73
228 640
Loop)
Winter
17.9
42
TABLE 3.7
 SPRING
RATESOF INDIVIDUAL RAIL TRACK COMPONENTS,
LUBER ( 196 1)
~ ~~ ~
Component
Softwood Sleeper W / m ) Wood
Hardwood Sleeper (kN/mm) 50 300
Steel Sleeper ( kN/mm ) 50
50
Concrete Sleeper (kN/mm)
50
Intermediate Layer 50
Sleeper Ballast 50 Overall Spring Rate (without ballast and formation) 30 Overall Spring Rate
20
 500 50  150
300 50
 500 500
300

500
 500 20000 300
2000 4000 8000 300
50
 110  80
40
20
 290
130
50
20

310
50
30
430
170
 180
TABLE 3.8

RESULTS OF GERMAN RAILWAYS (DB) TRACK SPRING RATE JEASUREMENTS, BIRMANN (19651966) Track Spring Rate (kN/mm)
Subgrade Type
on Track Track soil on Track on Track on Track on ent Most
clay gravel rock frozen ballast and formation
 15 15  20 20  60 30  40 80  160
5
30
43
TABLE 3.9
 VALUES
OF
THE
COEFFICIENT OF SUBGRADE REACTIONFOR
A
TRACK WITH A SLEEPER SPACING OF 630mm, AND ASSUMED EFFECTIVE SLEEPER BEARING AREA OF 5.2 x 105 mm2 (EISENMANN 196933) Subgrade Type Coefficient of Subgrade Reaction
106 (kN/n~m~)
Very poor subsoil (marshy soil, fine grained sand) 20 soil) Poor (cohesive subsoil Good 49
98
soil
CALCULATION
OF
THE
RAIL
BENDING
STRESS AT THE
BASE OF THE RAIL
Using the beam on an elastic foundation analysis under the action of a single design wheel load rail the bending moment M (kNm) at a distanceX (m) from the load point is
M ,
where P

BX
‘se
4.8
(cos ~x
 sin
B X ),
(3.32)
B
k E I
single wheel load (kN) 0.25  (XI) I = track modulus (MPa) , = Youngs Modulus of the rail steel (MPa), and = rail moment of inertia (mm41 .
=
A master diagram (Hay, 1953) has been developed, for the case o a single point load, which relates rail the bending moment under the load point Mo(kNm), to the rail bending at moment any other
location (Figure 3.12). The rail bending moment under the load point is the maximum for the single load case and is expressed as

48.
44
P
(3.33)
The distance x,(m) of zero rail and is given by
(Figure 3.12) is that distance to the position moment from the of point load application
bending
(3.34) For the will be actual track loading conditions rail the at any point moments caused by
subjected to a
combinationof bending
the interaction of adjacent wheelloads. The total length of this zone of interaction is approximately a distance 6x1, to the left and right of the load point, for the caseof a single point load (Figure 3.12). Using the principle of superposition the rail bending moment M(kNm) under a particular wheel including the rail bending moments due to the interaction of adjacent wheels is (3.35)
0, 1, 2 ..., = number of adjacent wheels in the interaction length (Figure 3.12). i = 0 refers to the reference wheel, distance to the adjacent wheels from the reference
where i
= =
xi
whee 1(l) (Figure 3.13) , PXi
1
= magnitude of the impact factored wheel loads at =
distances xi from the reference wheel, and expression for calculating the rail bending moment coefficient for any location using the beam on an elastic foundation analysis (2).
It can clearly be seen that the naximum rail bending moment depends to a large extent upon the axle spacing.
(1) Wheels at distances Xi>6Xl are not included, for the case
of the reference wheel X , = 0. (2) For a seneral case the exaression can be written as: BXi AXi = (COS Xisin X1 .) 46
e
45
Rail
l
I
E
I
I
1
Relative
value o f track
Y
Values of Bending
0
X I
2x1
3x1
4x1
Load Point
5x1
6x1
7x1
8x1
Distance Along Rail F r o m
Figure 3 . 1 2 Master diagram for moments, pressure intensity and rail depression under a singlewheel load (Hay 1953)
46
Typical W a g o n s
am JL
3
2
1
0 1
4
Rail
5 Wheel No.
\rrrf;
Rail P3
Reference Wheel Load
P4
PO
P1
X4
4 H
6x1
6x1
:1"I'"
P4
P5
41 i
v
~
l
12x1
S 5
*
Total Interaction Length
+l
Note: X1<
X2 < X 3 < X4 < 6x1
Figure 3.13 General load interactiondiagram for the calculation o f m a x i m u m rail bending m o m e n t and deflection
47
Some
railway
operators
restrict
the
length
of
the
vehicle
whe
base to less than 6x1, thereby reducing the occurrence of numerous fully unloaded cycles in the rail during the passage a train. of This is an important consideration in the fatigue life a rail. of The maximum rail 'stress at the centre of the rail base u b (MPa) is readily calculated using simple applied mechanics, can and be stated as
' b where
 Mm 
106
I
(3.37)
zO
& $ , , =
Zo
maximum rail bending moment (kNm) from Equation 3.35, and section modulus of the rail relative to the rail 3 base (mm 1 . rail sizes as
=
The section properties of current Australian manufactured by BHP are presented in Table 3.10. Allowable rail bending stress at the rail base
The following are methods currently used by railway organisations for evaluating the maximum allowable rail bending stress at the rail base. The General Method as used by AREA: The AREA (1973) recommends
that the acceptable rail stress for continuous welded rail be established at the rail base (Robnett al. et 1975). The current procedures limit the allowable rail bending stress in the rail base to implicitly avoid fatigue cracking. Clarke (1957) has suggested that the value of the allowable rail bending stress
should not exceed 50 per cent of the rail yield stress, uY. Although the value of the design load P contains the amplification effectof the impact factor which implicitly includes the effects of locomotives, track condition etc., the allowable rail bending stress as determined by the general method makes further reductions for these and other factors.
48
'I'AljLIS . . .
3.10  SECTION PROPERTIES OF AlJSTRALIAN M I L SIZES (BHP, 19T.S)___
M ~ S S < t n d l;octi.onof
i<ai
Area
of
L
Area of
Area of
'l'utal Ileight
Area
Total
Ilead
Web Foot
of llcight Neutral Inertia Head
Width Width Vertical of of Moment of Foot
1loriznnt.al. Sr:ction Moment of Modulus Inertia
Section
Modulus (Foot)
(a)
N r : W
Austra1,ian rail. section.
The
allowable
rail
bending
stress
should
be
sufficiently
below
the elastic limit (or yield stress) of the rail steel in to ord account forany variability in the rail support, the wheel loading or other existing service conditions, which may result excessive rail bending stresses in the inelastic domain or in the worse case, actual rail fracture. The general approach for calculating the allowable bending stress in the rail, uall(MPa) (Hay, 1953), relies on the application of various factors of safety as follows, consequently (3.38)
where
Y ut A
B
C
= = = = = =
yield stress of the rail steel (MPa), temperature induced stress in the rail (MPa), stress factor to account for lateral bending of the rail, stress factor to account for track conditions, stress factor to account for rail wear and corrosion, and stress factor to account for unbalanced superelevation of track.
D
According to 14agee (1965) the recommended values of these stress reduction factors can be explained as follows:
Lateral Bending: Due to the wheel loading having a horizontal component that produces bending of the rail in a horizontal plane, a lateral bending stress is produced in the base of the rail which is additive to the vertical bending stress one side or the other. Examination was made of Talbot's reports and a value of 20 per cent was considered adequate for lateral bending at all speeds, (Talbot 19181934). Track Condition: Due to the occurrence of mechanically worn, deteriorated, or low sleepers, increased rail bending stresses can be induced. The standard of maintenance or the attention which is given to proper sleeper support will of course determine the extent to which rail stresses will be
50
on
increased as a result of track condition. After examination of the Special Committee's reports (Talbot 19181934), a factorof 2 5 per cent was considered adequate to provide for the effect of track condition at all speeds track and 35 per cent in branch line track. in mainline
Temperature Stresses: In jointed rail track, the temperature stresses of concern will be the tensile stresses in winter due to joint bar resistance rail to slippage augmented to some extent by rail anchor restraints. Available data indicate that for main line track, temperature induced rail tensile stress will not exceed 69 "Pa and for branch line track, 34.5 MPa. With continuous welded rail, an allowance of 138 MPa is recommended in order
0
to
account the
for
the
rail droppingto a temperature of 38 C below temperature. Rail Wear:
laying
On the outer rail of curves an allowance should
be made for reduction in strength due loss to of area by wheel flange wear and corrosion.A study of the section modulus about the base of typical curve worn sections indicates thatan allowance of 15 per cent is adequate. Unbalanced Elevation: On curves, AREA recommendations limit the speed of operation to account for a probably 75 of unbalanced elevation. For a neight of the centre of gravity of a vehicle of 2.13 m, this would result in an increase in wheel load on the outerrail of 15 per cent. A comparisonof other recornended values of the stress reduction factors is presented in Table 3.11. This table outlines the stress factors recommended by Hay (1953)" Clarke (1957) and Magee (19651971). Magee includes the effects of both jointed and continuous welded track, whereas Hay and Clarkeanlist additional stress factor to account for the effects of the locomotive driving wheels. The additional rail stresses caused by the impact of rolling wheel flats are not considered. The calculation 51
of the temperature stress ot used in Equation 3.38 will
be
investigated in detail in a later section. It would appear from inspection thatthe majority of the stress safety factors are related to the maintenance condition of the track, and the magnitudes are based entirely upon the researchers' own judgement TABLE 3.11
 CRITERIA
Stress
FOR CALCULATING THE ALLOWABLE BENDING STRESS IN THE RAIL AT THE RAIL BASE (KOBNETT et al. 1975)
ResearcherFactor
yee Clarke Description Hay Symbol
OY U t
Lateral A Track Condition B
Rail Yield Strength (MPa) Temperature Stress (MPa) 48(f) 48(f) 34.5(a)
413
413
~
483
69 13 8
!E;
Bending
15 per 15 per 20 per cent cent cent
25 per 25 per cent cent cent 25 per (d)
(e) 35 per cent 15 per cent 15 per
C
Rail Wear and Corrosion
10 per 10 per cent cent
Unbalanced D Superelevation Locomotive
1520 25 per cent cent cent
5 per 5 per cent cent

Note: (a) Branch Line Jointed Track (b) Main Line Jointed Track


~
(c) (d) (e) (f)
Continuous Welded Rail Main Line Track Branch Line Track Jointed Track.
52
Wheel
flatsare
a
form of wheel
defect
resulting
from
the
wheels
sliding along the rails during inefficient braking. These flats apply dangerous inpacts on the rails, wbile the wheels roll along. Further dangers for the rail are due to rail or brake block material being built on up the wheel surface during slow slippage of wheels (material weldec onto the tread wheel surface). The danger to which the rails and the track stability are exposed in this case, and whicn also affects the of life the wheel sets and the vehicles in general, has been considered so great, that a European agreement limits the versine of 'depth of wheel flats to 1 mm, and the flat lengthto not more than 85 mm (ORE: 1965). However it is often noted in practice that the size of a wheel flat is considerably greater than the limits given above. The A m A (1952) in association with the Association of American Railroads (AAR) have carried out experiments to determine the magnitude of the rail bending stresses generated by wheel flats. The following general conclusions were drawn:
.
the rail stresses caused by wheel flats increase ragidly with speed reaching a maximum value at 30 km/h decrease to a minimum at 60 km/h and then begin to rise up again to140 km/h, although not reaching the value 30 atkm/h
.
the bending stress in the rail is proportional to the length of the wheel flat the then current AAR rnaxirr,um wheel flat length limit of 2.5 inches, 36 mm, did not produce excessive rail stress, but stresses were however 100 to 150 per cent greater than without the flat spot
.
. .
rail stresses increased with increasing wheel load the depth rather than the length of the wheel flat had a much greater bearingon t ' r e impact on the rail.
53
Schramm (1961) supports these views and states that the rail bending moment which is attributable to a wheel flat is largest at speedsof approximately 30km/h and is always dependent upon the depth of the wheel flat. The rail bending moment Mf(kNm) attributable to a wheel flat can be calculated from the following equation Mf where
PS
= = =
(15.4 + 0.11 PS) df0.5 ,
(3.39)
df
static wheel load in the range of 50 to 100 kN depth of wheel flat (mm); commonly occurring depths are 2 mm.
It is not readily apparent whetherthe effects of wheel flats are implied in the value of the impact factor used in the calculation of the design 3.11. The German Method: At a specified constant tensile stress ut wheelload or in the locomotive factor in Table
due to temperature and internal stresses, the allowable tensile stress uall based upon fatigue considerations can be determined for a particular rail steel material. Eisenmann (1969a,b,c) uses a fatigue diagram previously developed by Smith (1942) which reduces the allowable fatigue bending stress at the centre of the rail base as the apparent constant longitudinal rail stresses varies with fluctuationsin the rail temperature. This is similar to the SouthAfrican approach of analysing rail fatigue and referred to by Westrail (1976).
Values of the allowable rail tensile stress developed for Centra European conditions are listed in Table 3.12. The calculationof the constant rail temperature stress U t will be investigated in detail in a later section.
54
TABLZ 3.12
 ALLOWABLE
FATIGUE
BENDING STRESS AT THE CENTRE OF THE
RAIL BASE (EISENMANN 1969b) Rail Strength Allowable fatigue bending stress at the centre of g ult all (MPa) Jointed Xail cr = 78 (MPa) CWR ‘a’ 3 t = 176 (MPa)
6 86
U base rail the (MPa)
;
0. 33JUlt 0 360ult
882
=
313
0.3N ult
= =
225 274
(a) This constant temperature induced rail stress relates to .very low temperatures experienced under Central European conditions. For calculations of zt refer to p. 68 et seq.
Birmann (1968) consideres the effects of the lateral guiding forces on the magnitude of the tensile stresses in rail the base. According to laboratory tests and measurements in the track, the highest rail flange stress g R caused by vertical and simultaneously acting horizontal forces is approximately 1.4 to 1.6 times the stress at the centre of the rail flange,c b ’ urder vertical wheel load alone. Taking into account the longitudinal rail stress arising from temperature fluctuations the maximum combined stress in the rail is limited by (3.40) where
oall= allowable tensile stress in the rail (IIPaj,
U
Y
= =
yield stress of the particular rail material (MPa), base(MPa)and longitudinal stress in the rail arising from temper ature changes (MPa) .
ab
= calculated rail stress at the centre of the rail
ut
Having determined that the allowable bending stress at the centre of the rail base is not exceeded for a particular rail section, the rail is further analysed for its capacity to withstand the expected lateral loads in service.
55
CALCULATION OF THE
RAIL
BENDING
STRESS AT THE
LOWER
EDGE OF THE
RAIL HEAD
The magnitudeof the rail bending stress at the lower edge of the rail head (Figure 3.8 Point B ) is dependent upon the magnitude of the lateral guiding forces imposed on the rail head. According to Navier's hypothesis, an additional tensile bending stress has to be added to the usual tensile bending stress calculated at the lower edge of the railhead as a result of the discrete nonuniform shape of the rail profile and due to the fact that the wheel force acts to some extent in the horizontal direction (Eisenmann 1970a). eccentrically
a
Combining the applied rail head load due to the eccentric and skew wheel load application, (Figures 3.14), gives an equation for determining the stress at the lower edge of the rail head o K , i.e.,
aK =
(3.41)
Au
2 ,p
stress dueto vertical load applied centrally, additional bending stress caused aby centrally applied force due to the discontinuity effects at the railweb, additional bending stress caused by the torsional P, moment dueto the eccentricity of the axle load additional bending stress caused by the torsional moment due to the eccentricity of the guiding force H, bending stress due to the horizontal guiding force applied at the fulcrum. and oH are
The magnitudes of the terms up, n o l l A U ~ , ~ , calculated in the following manner.
The stress at the levelof the lower edge of the rail head up(MPa) due to an equivalent centrally located vertical design wheel load, P is
56
o +
(a)Vertical Load
P
1
M,=P.e
P
1
1
(b) Horizontal Load
(c) Skewed Load
+
M
2
+
S
Note: M , = H  h  P . e
Figure 3 . 14 Superposition of r a i l stresses caused by vertical and horizontal loads to obtain resultant r a i l stresses caused by a skewed load (ORE 1966)
24722/80"1 I
57
5P
where
Mm.hl. 106 ,
(3.42)
%
A1
=
= cl =
I
=
maximum rail bending moment (kNm) calculated from Equation 3.31, 3 = rail sectionparameter (mm I ' height of the position of the lower edge of the rail head above the neutral axis of the rail (mm), and 4 moment of inertia of the rail (mm 1 .

The additional bending stress O1(MPa) caused by the disturbance of the rail section where the web meets rail the head edge is calculated from the equation
A5
1
where
P
h
= = = = =
=
X p 1 0 3,
(3.43)
(kN), and
'
design wheel load
2
rail section parameter (mm2) ;
A 2 is given by A
2 where
1.5
r.
L
3 kn
(a2/alTl 0.25
I
I
(3.44)
3 b3.a.4.d
J
al
a2 b d
average rail head height (mm), height of rail head + web (mm), web thickness (mm).
(mm), and = mean rail head width
The additional bending stress AD2 (MPa) caused by the torsional moment, at the levelof the lower edge of the rail is calculated from the equation
A5
2
=
h3 MT.10 6 ,
(3.45)
58
where
= = = = =
=
applied torsional moment to the rail 3.14~) H.h P.e, lateral guiding force (kl?) ,
(kNm) (Figure

H
P h
e
l3
design wheel load (kN), distance between the position of the loading point of the lateral force and the fulcrum, (m) eccentricity of the vertical force (m), and rail section parameter ( m ~  ~ (the ) theoretical 1932.) solution is given by Timoshenko and Langer
=
For any rail section the applied torsional moment MT will be of simple twist and partially transmitted partially in the form by bending of the head and base of the rail, thus
where
M1
fif2
= =
rail section torsional resistance (St. Venant's torsion) , and the torsional bending resistance of the rail head and rail flange, (i.e. warping torsion).

From these considerations the value of the rail section parameter l 3 can be calculated neglecting the effect of the rail web, by A3 where
E G
 E.h 2.GI
~
u( 1 e " ' )
P
l+eaS
 2
,
(3.46)
e b
S
= = = = =
Young's Modulus of the rail steel (MPa), Shear Modulus of the rail steel (MPa), natural log base, mean width of rail head (m) ,
IP 1
P
distance between the rail supports (mm) (i.e. the sleeper spacing) 4 = polar moment of inertia of the rail (m), + I and = I p (flange) ' p (head) IP (web 1
+
G.1
=
P
F . I H (hH + hF) .hH
l 1
59
(3.47)
The distances hH(mm) and hF(mm) refer to the distances from the
of the rail neutral axis of the centroid respectively, and are defined by
head
and
rail
flange
hH
=
h t IH + IF , and
(3.48)
where
IH
= = =
=
ht
4) horizontal moment of inertiaof the rail head (mm horizontal momentof inertia of the rail flange 4 (mm ) distance between the rail head and rail flange centroids (mm) , hH + hF . (3.50)
The bending stress aH(MPa) at the
levelof the lower edge of the
rail head due to the lateral guiding force H(kN) applied at the fulcrum is
oH
h4.H.10 ,
3
where
X 4 is
x4 =
rail section parameter (mm2);
determinedby
(3.51)
where
zh =
horizontal bending resistance momentof the rail head, and is defined by
and I ~ , IF S , and b are as previously defined.
60
Calculated values of the rail section parameters xl, h 2 , x3, and x 4 for common AREA and German rail profiles are presented Table 3.13. The theoretically calculated value of the stress at the lower edge of the rail head is generally 10 per cent higher than the experimentally measured value. TABLE 3 .l3
in

PARAMETERS TO CALCULATE THE ADDITIONAL BENDING STRESS ON THE LOWER EDGE OF THE RAIL HEAD (EISENMANN 1970) AREA AREA AREA S49 ll5RE 132RE 140RE
57.5
115
~
Rail Weight Rail Section (kg/m) S54 54 98 128 108 S64 64
66 132
49 70 140
( Ib/yd)
~~ ~
Moment of Inertia 6 4 I (10 mm 1
~ ~~~ ~~~
27.20
35.80 40.88 18.9 20.73 32.52
Section Modulus rail base
about 356 262
1.51.5 1.7 2.0
zo
X1
A2
X3
474 441 ( 1 0 ~ ~ ~369 ) 240
( 1 0  ~ ~  ~2.2 )
3mm 2 ) (10
1.8
0.655 0.612 0.519 0.540 0.490 0.480
( 1 0  ~ ~  ~ 19.7 17.9 14.9 ) 16.0 14.0 13.0
h4
(10~~~1.30 ~)
1.06
1.00
1.60 1.40
1.00
Distance from fulcrum to upper sideof rail head h
(mm)
13 0
140
145
99 124 102
NOTE:
Calculated values are10 per cent higher than experimental.
61
Eisenmann (1970a) states that the flexural stresses in rail the head reach a considerable value only for comparatively high lateral forces. The experience gained from numerous intrack measurements has shown that locomotives initiate the higher values. The wheel sets of locomotives with a wheel load of 110 kN exert lateral forces of about 44 kN to 59 kN, while freight cars with a wheel load 90 ofkN deliver only 15 kN, assuming in both cases a degree of curvature of 6 degrees. (i.e. curves with a radius of approximately 300 m.) The number of repetitions of excessive stresses occurring at the lower edges of the rail head are small compared with the foot of the rail, and therefore a
,
fatigue failure is not expected to occur at the lower edge of the rail head. The tensile bending stress oK occurring at the lower S54, edge of the rail head in the case of German rail profiles UIC 54 and UIC 60 under the simultaneous application of vertical and horizontal loads has been calculated using the above analy Determining the rail bending stress at the centre of the rail flange ub (Equation 3.33) calculated for various centric wheel loads and values of track modulus, a ratio uK/ub can be plotted against various values of the vertical wheel load, the lateral guiding force and track modulus is and presented in Figure 3.15. It is readily apparent that for certain combinations of high
lateral guiding force and medium to high wheel loads, the magn of the tensile bending stress occurring at the lower edge of the railhead can be one to two times greater than rail the bending stress calculatedat the centre of the rail flange. CALCULATION OF THE VERTICAL DEFLECTION OF THE RAIL
Having determined that for the design lateral loadingrail the section under consideration satisfies the permissible limit of stress at the lower edge of the rail head,the rail is further analysed for its deflection performance under load. Using the beam on an elastic foundation analysis under the action of a single wheelload the rail from the load point is deflection y (mm) at a distanceof
X
62
2
H = 60
1
H = 40
0
H = 20
60 80 100 Wheel LoadP (kN1 S54 120 1 4 1 J
40
40
60
80 1 00 Wheel LoadP ( k N ) UI C 54
120
140
2
H = 60
1
H = 40
0
H = 20
60
4 0
80 Wheel
120
140
Legend:
" "
LEok
ok = Stresses at lower
i
Load P (kN) U!C 60 Ballast coefficient C = 10 kg!crn3 i.e. track modulus k = 40 M P a (approx.) C = 5 kg!'cm3 Ballast coefficient i.e.track modulus k = 20 M P a (approx.) edge o f rail head.
(Up+iliUl+aOi,,+AU,,~+UH)
ob = Bending stress incentre of rail foot
\
00
H = Lateral Guiding Force ( k N )
Figure 3 . 15 Ratio between the calculated bending stresses in rail head andrail flange of @K/db ( O R E 1966)
63
YX where
P
(3.52)
= = = = =
single wheel load (kN) ,
[+I~~~,
track modulus (MPa) , Young's Modulus of the rail (MPa), and 4 rail moment of inertia (mm 1.
Similar to the case of the calculation of the rail bending moment due to a single point load, a master diagram (Hay, 1953) has been developed which relates the vertical deflection of the rail under the load point yo(mm) to the vertical deflection of the railat any other location (Figure 3.12). For a single isolated load the maximum vertical rail deflection occurs directly beneath the loa point and can be expressed by (3.53) The distance contraflexure x2(m) from 3x1, is the that distance of to the position of rail and is by given (3.54)
point
load
application
x2 where
=
x1
is defined in Equation 3.34.
For the actual track loading condition the vertical at any position along the rail is the sum of the deflections causedby the interaction of wheel loads point. The total length of this zone of interaction is mately a distance 6x1 to the left and right of the
rail deflect vertical rai about this approxiload point,
for the caseof a single point load (Figure 3.12). Using the principle of superposition the maximum vertical deflection of the rail ym(mm) under a particular wheel load, including the vertical deflections due to the interaction of adjacent wheels is
64
L
i=Q ' x i where i
Bxir
(3.55)
=
0, 1, 2 ..., is the number of adjacent wheels in the interaction length commencing at the reference wheel (Figure3.12) , distance to the adjacent wheels from the reference wheel (1)I
magnitude of the impact factored wheel loads at distances Xi from the reference wheel, and expression for calculating the vertical deflection of the rail coefficient for any location using the (2) beam on an elastic foundation analysis Allowable vertical deflection of the rail
With most typical track structures, the rail acts as a continuous beam on an elastic foundation. The resulting deflection of the rail caused by the rolling wheel creates a "wave motion" in front and behind the wheel. The rail being fastened to the sleepers which are embedded in ballast, and restrained longitudinally either by rail anchors or by elastic fasteners, is thereby prevented from sliding about the track. The track therefore, resists the motions of the wave, and although vertical motion is observed, the longitudinal movement is counteracted by resistive forces of the track itself. The value of the theoretical rail deflection is, for a given rail size and axle loading, dependent upon the assumed track k modulus under one rail and the spacing between adjacent axles. It can clearly be seen that care must be exercised by the designerin the selectionof the assumed value of the track modulus if realistic valuesof rail deflection are to be determined.
(1) Wheels at the distances Xi<6X1 are cot included, for the
reference wheel X,
(2)
BX
= 8 . e
2k
=
0.
n 7r
i
24722/80l2
65
For typical track structures designed with light to medium rails the AREA has recommended that the value of track modulus to be used when determining the rail deflection, 13.8 be ! @ a (2000 psi). Theoretically the value of the track modulus cannot be assumed be proportional tothe rail size because it only relates theto support givento the rail by the sleepers (including rail pads and fasteners), the ballast and the subgrade.
t
Larger rails are designed predominantly to carry heavier loads which consequently demand the adoption of a heavier track structure (i.e. a reduced sleeper spacing in conjunction with a probable increase in the ballast depth). It is the combination of these factors in conjunction with a reduction in load the distribution due to a stiffer beam, that lead to larger values of track modulus when the rail size is increased. It would appear that the AREA (1973) manual recommendation 1975.22.3.15 of a maximum allowable vertical deflection 6.35 of mm (0.25 in) is based on a track modulus value of 13.8 MPa.
Lundgren et al. (1970) has incorporated this recommendation in his rail deflection limits which are based upon the capabilit the track to carry out its design task (Figure 3.16). LONGITUDINAL TEMPERATURE STRESSES INDUCED IN THE RAIL
In addition to longitudinal stresses created by the wave action, there are two other longitudinal stresses of equal or greater magnitude which affect the rail design. The first of these is
due to the forces exerted by the wheel of the locomotive when either accelerating or braking. This is a complex condition which is theoretically highly indeterminate and has not been measured intrack successfully. The second of these is due to thermal expansion and contraction caused by changes in the ra temperature from that experienced when initially laying the
66
0.0
1.27
2.54
3.81 7.62 6.35 5.08 8.89 11.43 10.16 M a x i m u m Track Deflection inmi
12.70
Legend:
A B
C D
Deflection range for track which w i l l last indefinitely. N o r m a l m a x i m udesirable m deflection for heavy track to give requisitecombination of flexibility and stiffness. Limit of desirable deflection for track of light construction (G50kgim). Weak or poorly maintained track which w i l l deteriorate quickly.
Note:
Values of deflection are exclusive of any looseness or play between rail and plate or plate and sleeper and represent deflections under load.
Figure 3 . 16 Track deflection criteria for durability (Lundgren et a1 1970)
67
track. The importance of knowing the magnitude of the longitudina temperature stresses in the rail is that these stresses significantly alter the allowable rail bending stress. Current American practice is to construct track made from flash butt welded rail of lengths up 490 to m. This is laid on timber sleepers fitted with steel bearing plates. The lengths of rail are joined either by thermit type welds or by adhesive bonding and are held down to the sleeper by the use of dog spikes. The longitudinal actionof the track is resisted by the rail anchors which are boxed on either side of the rail. Patterns
of anchorage vary, but usually every other sleeper is boxed a in long stretches between rail joints every third sleeper is bo anchored. At breaks in therail such as insulated joints all sleepers are box anchored from three to six rail lengths both sides of the break. The structure works on the basis that the longitudinal force is transferred from rail the to the anchor and the ballast resists the load distributed from the sleeper. Vertical rail uplift caused by wave motionis not constrained, the dog spike being incompletely driven home so as to allow the rail to "breathe" upward before and after the wheel. This leads to a rather loose arrangement where all visible movement is done by the rail. The AREA method for calculating the longitudinal stresses
The general approach suggested by the AREA (Robnett et al. 1975) and as outlined in Table 3.11 is to adopt the value of the increase in the rail stress due to temperature changes as one the following values:
a
t
=
138 MPa : CWR track 69 MPa : jointed mainline track 34.5 MPa: jointed branch linetrack.
ut
ut
= =
68
The maximum end movement of the rail Sr(mm) from the formula (Eisenmann 1970a)
can
be
calculated
(F J ~2s ) 103 2 . R Ar E 1
where F

(3.57)
=
total force required to fully restrain the rail against any rail movement due to a temperature variation from the rail temperature at laying Ar
t)
0
1 5
tE.10 3 (kN)t
cross sectional area of the rail ( m m2 coefficient of thermal
1.15
10~ (Oc5 ,
expansionof rail
steel
Young's Modulus of the rail steel (MPa) , maximum temperature deviation from base
0
laying
temperature ( c) , base laying temperature of the rail (OC), rail joint restraint force [assume Jr = 0 for track with dog spike rail fasteners), sleeper spacing (mm), and average sleeper resistance movement (kN/Sleeper/Rail) . The value to rail longitudinal
the maximum end movement of the rail Of
r(max) is
usually assumed as 9.5(mm) and Equation (3.56) is rearranged to solve for the minimum longitudinal rail restraint Ro (min) ( kN/ sleeper/rail) required by therail fastener,i.e.,
(min) The restraint need not movement ballast, sleeper
 F2S.10 26r(max) A
provided
3
r E' the rail and the sleeper,
(3.58)
between
however,
exceed the ability of the ballast section to restrain of sleepers longitudinally in the ballast. For gravel the restraining force provided by the ballast to the does not exceed 5.4 kN (AREA 1975) .
69
The
German
Method
for
calculating
the
longitudinal
stresses
(1961) makes a When considering jointed rail track, Schramm
distinction between the occurrence of ''short a rail" and a "long rail". The former being defined for the conditions where the rail joint gaps are not closed when the upper temperature lim are reached and the maximum gap width is not achieved at lower temperature limit. Whereas the latter is defined for the
the
conditions where the rail joint gaps are closed when reaching upper temperature limit, or the maximum gap width is reached at the lower temperature limit. If the rails are prevented from moving lengthwise when they approach the limits of the tempera range thena.condition similar to continuously welded rail is reached. The following formulae have been suggested by,Schramm for use in the calculation of the increase in longitudinal rail stress caused by temperature changes: (a) CWR Track: The increase in the rail stressa t (MPa) due to a temperature variation from the temperature at laying can be calculated by
where tl to "t E
=
=
= =
the upper temperature limit of the rail (OC), the base temperature of the rail (OC), coefficient of thermal expansion ( O C ' ) , and Young's Modulus of the rail steel (MPa).
This maximum rail stress occurs in the central portion of
CWK and is constant
for
any
particular
temperature
differ
ential. (b) Jointed Rail Track: For the case of jointed rail the longitudinal rail force can be transferred from rail
to
r
via friction between the fish plates. According to Schramm
70
the
mean
force
which
can be d2veloped at the
fishplates
is
of the order of50 kN. The length of rail over which the temperature stress increases to a naximum value is also
called the breathing length. Over this length the cumulative value of rail end expansion or contraction is said to occur. (C) "Short Rail" Track: Such track is characterisedby &e fact
that, within the maximum texperature range the rail temperature at which the rail gap closes t exceeds the upper operating g rail temperature tl, it >tl). For this condition the
9
maximum longitudinal rail stress st(MPa) occursat the centre of the rail length, i.e.,
Ot where

2 F .
WO Er I 3 2  1 0 Ar
+
(3.60)
F . W ,
= = = =
Lr
A ,
force transmitted fromrail to rail via friction between fishplates (kN), track resistance t c , longitudinal movement of the rail per length of rail (kN/m), (depends upon the fastener rigidity) , length of jointed rail between adjacent fishplates (m) and cross sectional area of the rail (mmL ). states that under German conditions the case of
Schramm
"short rail" trackis only academic as in actual practice the jointed rail track always behaves as a rail" "long track. (d) "Long Rail" Track: This track is claracterisedby the fact
that within the upper temperature range rail tne tenperature at which the rail gap closes t is less than or equal to the 9 upper operating rail temperature t;, (t . j r tl) . For this condition the maximum compressive lonqitcdinal rail stress ut(MPa) occurs at the centre of the rail length and is

independent of the fishplate friction Fo.
Consequently,
71
(3.61)
where R . = rail joint gap (mm), and g to, tl, E, W O and ir are as previously defined in "t' Equations 3.59 and 3.60. Values of R and W O for German operating conditions are presented g in Tables 3.14 and 3.15.
TABLE 3.14
 RAIL
JOINT
GAP
LENGTHS FOR PARTICULAR
LENGTHSOF RAIL,
GERMAN CONDITIONS (SCHRAI'IM 1961)
Length Gap Joint Rail Length Rail
,Q"(m)
~~~ ~
R _ (mm)
15 30 45 60
5
9 11 14
TABLE 3.15
 TYPICAL
VALUES FOR THE TRACK RESISTANCE TO LONGITUD MOVEMENT OF THE RAIL,WO PER (METRE) OF RAIL LENGTH FOR VARIOUS SLEEPER TYPES (SCRAMM 1961)
Type
Sleeper Timber Concrete Steel
wo
(kN/m)
3.9 4.9
5.9
For "long rail" track the "breathing length", Rb(m) can be calculated from
R
b
72
(3.62)
If the length of the breathing Lb becomes smaller than ar/2 then no longitudinal movementcan take place at the centre of the raiL, and the characteristics of CWR are exhibited. This occurs only if the length of the joint gapz
L rLWO. 103
9
is less than
2EAr Therefore Equation3.61 is only valid for rail joint at the time of laying where .L ,is greater than
9
p.
(3.63) gap lengths,
r
2wo .l0 3
2EAr STRESS CONSIDERATIONS
(3.64)
WHEEL
TORAIL CONTACT
High valuesof contact stress and shear stress are generated at the wheel/rail contact zone (Figure 3.8, point C ) , due to heavy vehicle wheel loads. Because of the ever changing demands of railway freight operations, it is important to determine the maximum wheel loads that a particular rail (and or track) can carry before irreversible rail head damage occurs. Therefore the calculation of the applied force level at which rail head shelling will occur is of fundamental and problems of a similar nature importance. The following will review nethods that have been developed to place limits upon the vehicle wheel loads. These include the simple P/D ratio, applied Hertzian theory and shakedown considerations.
l
The P/D ratio The ratio ofthe static gross wheel load P to the wheel diameter D, i.e. P/D, can be used as an indication of the magnitude of the contact pressure occurring between the rail head and the wheel. With current railway satisfactory. operations certain P/D limits have proven
24722/8013
73
In 1958 British Rail limited the ratio of the wheel load P (tonnes) to wheel diameter D to (m) 9.2 t/m (90.3 kN/m) . In for 1963 this value was increased to 12.5 t/m (122.6 kN/m) trailing wheels and 10.8 t/m (105.9 kN/m) for driving wheels (Koffmann and Fairweather 1975). Birman, quoted in the ISCOR(l) Report (Taute et al. 1971), states
U
ult = (107.9117 kN/m) caused no contact pressure problems between the rail and the wheel. Whereas wheels with P/D ratios of 1417 t/m, (137.3166.8 kN) caused considerable damage the to rail head and spalling occurred in the wheels. American practice is to establish general limits for the maximum wheel load P(t). Given (Koffmann and Fairweather 1975)
P
that for a rail steel an with ultimate tensile strength 90 kg/mm2 883 MPa wheels with P/D ratios of 1112 (t/m)
=
3 3 ( ~ ) " ~ 18.4
' 
(3.65) becomes, (3.66)
the
equation P/D
for
the P/D ratio 18.4 D *
33. = Do. 5 
P/D
ratios of 16
 18.5 t/m
(157
 181.5
kNm) are attained
in service. The P/D ratios currently used by various railway operators and the P/D ratios recommended by various railway organisations are presented in Table 3.16. It can clearly be seen that the Australian mining company railway operators their design wheel load upon North American practice.
base
Although the simple P/D ratio provides a convenient indicator of whether contact stress problems are likely to occur, it oversimplifies the situation and should only be used to determine whether more detailed analysis is required. The limits by set British Rail (Koffmann and Fairweather 1975) or by Birmann (Taute et al. 1971) provide conservative levels for this purpose.
(1) ISCOR: South African Iron & Steel Corporation. 74
TABLE 3.16

P/D RATIOS CURRENTLY 1971, READ 1972)
USEDBY RAILWAYS
OPERATORS
AND
RECOMMENDED BY RAILWAY ORGANISATIONS (TAUTE et al.
Railway Organisation Operator or British Railways (BR) French Railways (SNCF) Japanese Railways (JNR) South African Railways (SAR) Australian State Railways AAR Recommendation Hamersly Iron Railways Quebec Cartier Lamco J V QuebecNorth ShoreLabrador
P/D Value (kN/m) (tonnes/m) 10.8 9.0 9.2 11.4 9.2 14.7 16.3

12.5 10.0
105.9 88.3 90.3
111.8 90.3

122.6 98.1 105.9
 10.8
17.1 18.3 16.9
144.2 159.9 167.8 179.5 165.8
The calculation of the wheel/rail contact stresses and maximum allowable shear stress in the rail head
the
An examination of the stress components in the rail head shows that in the immediate vicinity of the contact point between the wheel and the rail the principal stresses are very high (Eisenmann
1970) (Figure 3.17), with the following assumed stress condition occurring:
Although the extreme values of the conpressive stresses exceed the ultimate tensile stress of tne steel, no failure occurs because the shear stresses vanish at the surface, i.e., (3.68)
The
compressive
stresses
lead to plastification and thus to
hardening of the steel in the top zones of the rail head. With increasing depth below the rail surface the major principal stress o1 in the direction of the applied load decreases slowly whilst the minor principal stresses a2, u 3 decrease very rapidly. As a resultof this principal stress differential with increasing rail head depth a maximum value of shear stress, amounting to approximately 30 per cent of the specific compressive contact at a depth corresponding to half stress uc at the surface, occurs the contact length (Figure 3.17)
.
Eisenmann (1970a)in an analysis of contact and shear stress levels assumes that the contact pressure is uniformly distributed over the contact area and that the rail and the canwheel be represented by a plane and a cylinder. The contact length 2a(mm) is derived from Hertz‘s formula andcan be calculated by 2a where
R E
=
= =
3.04
r
P R . ~ O O~o
1
5
,
(3.69)
P
vertical wheel load (kN), wheel radius (mm), Young‘s Modulus of rail steel (MPa), and breadth of contact area (mm); Eisenmann adopts the value of 2b = 12 mm.
=
=
2b
Due to the compressive stress being assumed to be uniformly distributed over the contact surface, the contact pressure a (MPa) can be calculated by C
aC

P.10
3 (3.70)
2a 2b
When the ratio of the contact pressure, o c to the rail yield 2.0 the useful life of the rail strength, ay is such thatoc/oy will be limited only by wear (Taute et al. 1971). Neglecting the effects of work hardening, total plasticity at the contact surface occurs whenac/ay 2 3.0. From fatigue considerations the allowable contact pressure a (1970a) is ‘all 76
(MPa) according to Eisenmann
Minor Stress
1.0 0.8 0 . 6
03
Major Stress
0 1
0.4
0.2
1.0 0.8
0.6
0.4
0.2
Depth
03 Distribution
Depth
9 Distribution
Sheer Stress
oc
0c
7
0.4
0.3 0.2
0.1 0
a 2a 3a 4a
5a 6a Depth
L Distribution
DC
Note: Contact Pressure U,
= 2a2b
~
P
I
Principat Axes
Figure 3 . 17 Half space with a strip loadand the resulting stress distribution with depth (Eisenmann 1970a)
77
(3.71)
where ault
=
ultimate tensile strength of rail (MPa).
As previously mentioned the maximum value of the shear stress, T,
is about30 per cent of the contact stress u C at the contact surface, and this value occurs at a depth of a (half the length of the contact surface). Thus the maxinum shear stress in the rail headT ~ ~ ~ ( can M P ~ be) calculated by
max
=
o.3uc
.
l
(3.72)
The hypothesisof shear strain energy applied for the condition of two principal compressive stresses u1 and u 3 gives for the rail and the allowable relation between the allowable shear stress normal stress uc all
(3.73)
Substituting
Equation3.71 into
Equation3.73 yields
the
maximum
allowable shear stress in the rail head T ~ ~ ~ ( from M P ~ fatigue ) considerations: where
0.3 *ult' (3.74)
occur A special type of fatigue failure, called shelling, may if This the rail head shear stress exceeds its allowable value. failure startsat rail head depths of 5 7.5 mm, where the shear stress reaches a maximum. The rail yeild stress can be exceeded by the occasional occurrence of extremely high wheel loads and the resulting shear stress moves the material into th

elastoplastic
domain.
Combining Equations 3.69, 3.70 and 3.72 and assuming that the breadth of the contact surface, 2b equals 12 mm,
78
can be derived (Eisenmann 1970a1, where R P
= =
wheel radius (mm), and wheel load (kN).
The results of this formula for wheel radii ranging from 300600 mm (1224 in) are sufficiently accurate compared with those for an elliptical contact surface region. The maximum shear stress in the rail head as a function of wheel radiusand wheel load has been plotted in Figure 3.18. Allowable limitsof the maximum shear stress in the rail head for various rail strengths are also shown. The problem of excessive maximum shear stress in the rail head can be solved by:
.
.
.
reducing the wheel load increasing the wheel radius increasing the yield strength of the rail,
U
Y
.
Once the shear strength of the rail steel, due to local contact stress, is exceeded the effect of an increase in the rail mass to prevent yielding from occurring will be of little value. Rail quality, axle load and wheel diameter a:Jst therefore be mutually adapted, and the rails should not just be regarded as carriers. The results of Eisenmann's analysis nay be adoptec? as a more P/D ratio. The sophisticated limiting criterion than the simple recommended limits draw on the experience of the D.B. with their wheel and rail profile combinations azd are recommended to avoid the occurrence of shelling of the rail head. At higher vertical loads, or in curves where high shear forces occur at the wheel/ rail contact interface, gross plastic flow nay be evident and further restrictions need to be placed on load levels.
79
400
9
Allowable Sheer Stress Limit
Wheel 100 0.30
Legend:
I
Load P = 40 k N
I
I 0.40
I
I
0.50
0.60
Wheel Radius r (m)
Permissible Shear Stresses Plotted As
(1) S t e e l uUlt= 685
Mpa Mpa
(2)S t e e l uUlt= 880 Mpa
(3)S t e e l uUlt= 1075
Figure 3 . 18 M a x i m u m shear stress in the railhead as a function of wheel radius and wheel load (Eisenmann 1970a)
80
Yield
and
shakedown
in
Dlainstrain
rollins
contact
The effective loading on the interface between the wheel and ihe rail head is both normal, due to vertical wheel loading, and tangential, due to driving and braking tractions. Desirable wheel loadings when transmitted to the rail head through the elliptical contact area, (applicable to rails with radiused rail heads) frequently exceed the yield limit of the contacting materials. Hence, the resulting surface plastic flow coupled with wear processes acts to flatten out the contact area. The contact surface can be approximated as bounded by a rectangle W based upon the assumption of contact of length 2a and breadth between a plane (rail) and a cylinder (wheel). The length, 2a (mm) canbe calculated from Hertz's formula, (Smith and Liu 1953) i.e. ,
2a where P R
W
= = = =
3 .l9
P(l
WE
V
2)R
0.5
I
(3.76)
wheel load (kN), wheel radius (mm) contact width (mm) , Poisson's Ratio and Young's Modulus (0.3 and 207 ; GPa for steel)
v,E =
with the breadth W treated as a statistical variable, assumed to be normally distributed, which depends on the conditions of wear of the wheel and rail. The corresponding maximum normal stressU ( " a ) aPP surface is (Mair 1974) (Figure 3.19) on the contact
aPP

2P
ra W 103
.
(3.771
It is usually assumed that the tangential traction q (YPa) is directly related to the normal Herzian pressure at all points the contact surface , with a maximum value
on
24722/8014
81
.
Figure 3 . 19 ( a) Schematic contact area forn e w wheel
and rail
Uapp
= S 2P
L..
F"
Figure 3 . 19 (b) Contact stress at shakedown
Contact surfacebetween rail and
82
wheel
(Mair 1974)
(3.78)
where
p
=
coefficient of contact friction.
In the absence of residual stresses the limiting elastic loading may be associated with either the maximum principal shear stress (Tresca) or maximum octahedral shear stress (von Xises) yield criteria (Mair 1974). Mathematically, these state that yield will occur when, either
' max

1 2 (u1
 u3)
=
k (Tresca k
=
oY/2)
(3.79)
(von Mises where k
k
= ./J3)
(3.80)
=
yieldstress
in simpleshear.
Extension of the rolling contact elastic analysis to the postyield condition has been carried out by Johnson and Jefferies (1963) for an elasticplastic material. Beyond initial yield the maximum contact load which can be sustained by a body in rolling contact, with limited plastic deformation, is known as the shakedown limit. Adopting the von Mises yield criterion, the effect of tangential force on the limit of elastic behaviour and on the shakedown limit is shown in Figure 3.20. For a prescribed tangential force/normal force ratio the limiting value of normal force, or alternatively the material strength to sustain a known normal stress, uall, is obtain from the vertical axis. Since rail defects due to gross plastic deformation are the result of cumulative damage, it is not necessary to preclude the exceedance of the shakedown limit by the applied stress, but it is necessary to ensure that the frequency with which it occurs is sufficiently small such that rail replacement takes place for other reasons prior to the deveiopment of an unacceptable degree
83
of deformation. Diagrammatically, this means that the overlap
between the distributions of applied contact stress, U and aPP" the maximum permissible contact stress uall, has to be limited to a predetermined level (Figure3.21). Expressed mathematically, R, against yield it is required to determine the reliability, such that the contact stress between the wheel and rail will exceed the shakedown limit of the material, where (Mair and Groenhout 1975, 1978)
(3.81)
However, the analysis is more readily accomplished by caiculating the probability of yield, Q, (represented by the region of overlap) given by Q
=
l  R
(3.83)
in which Fall(z)
=
allowable stress cumulative probability distribution function,
aPP
(z)
=
applied stress probability density €unction.
Mair and Groenhout(1975) take the reliability parameter to abe function of the following variables each of which is assumed to be a random variablewith a normal distribution:
.
.
. .
the the the the
rail material yield strength vertical wheel load contact area tangential forceon the rail head.
84
Tresca and V o n Mises Shakedown Limit
Tresca Elastic Shakedown
2V o n Mises Shakedown Limit
l
I
I
I
I
I
0
0 . 1 0.2 0.3 0.4 0.5 Tangential/Norrnal Load Ratio T/P =
0.6
Figure 3.20 Yield and shakedown limits in plane strain with combined normal and tangential loading (Johnson and Jefferies 1963)
Allowable Stress Reliability R = Prob (ua,, > uapp,) R . = Acceptable Reliability
Applied Stress
OaPP.
%l
Track of Acceptable Performance %Measure of Yield Probability
Figure 3 . 2 1 Schematic representation of interaction between applied and allowable contact stress values (Mair and Groenhout 1975)
85
TO obtain
an acceptable value of reliability, Ro, the parameters to tangent track were substituted into Equation 3.83,
relating
this approach being based upon the observation that gross plast flow (e.g. corrugations) is almost entirely limited to curved track. In particular, the tengential force in tangent track was assumed (conservatively) to be zero and the yield stress of standard carbon rail steel was used. On the above basisit can be shown that Fall (2)
 FU 
zm
 m
(3.84)
and
2 2 2z(x mwsp +
A ~
2 2
~
1S
~
Z (A%
fapp (2) =
J2 (A2S2 +
P
2 4 3 swz )
I

 mwz
1
2 2 224 2(A sp + swz
]
(3.85)

where F ,
 the cumulative distribution function of the standard
normal variable deviation, with mean zero and unit standard mean of subscripted standard deviation 2.31 Y, variable, of subscripted
variable,
l/(llJ) I material yield stress (?Uy)
I
wheel/rail coefficient of contact friction, vertical wheel load, wheel/rail contact width, 2 0.16E/ ( 1V R r wheel radius, and Poisson's ratio and Young's modulus. In most railway applications the operating parameters have set and the available means of adjustment to overcome rail deformation problems are usually limited to the replacementof the rail steel. On this basis Figure 3.22 draws on the experience
86
b
1200
l
A l
5 a
>
m
1000
t
R = 0.43m
0
R = 0.48m
5 m E
; i
800
I
R = O.43m R = 0.48111
400 I
l
120 140
180 160 200
100
Wheel Load (kN)
Yield Strength Requirements
1
1
R = 0.38m R = 0.43m R = 0.48m
I
800
R = 0.38m R = 0.43rn
R = 0.48m
4
100
I
120 160 180 Wheel Load (kN)
140
200
Ultimate Tensile Strength Requirements
Note: Reliability is 92.0%
Figure 3.22 Rail strength requirements to avoid corrugation
87
Newman Mining C O to establish recommended mean 0.2 per cent stress levels for systems having alternative combinations mean wheel load distribution and wheel diameter (Mair, Jupp and Groenhout 1978). The data presented can be used to prevent gross plastic flow (leading to corrugation). However, it cannot
of the Mt
be presumed that adoption of the above strength levels will compensate for poor standards of track maintenance nor the presence of excessive vertical joint misalignment, both of whic will the contribute Mt Newman to more severe conditions than those present on System.
The correspondence between the yield stress and the ultimate tensile strengthof the rail material for a range of as rolled alloy rail steels was obtained froma range can be expressed by (Figure 3.23) of test results and
(3.86)
or
0
ult
 m Y
of Figure 3.22 can be used as a guide to the
2080 uT,
(3.87)
The
data
selection
of required strength levels, however, for an accurate estimate the full reliability analysis should be used with the actual operating conditions.
LATERAL TRACK
STABILITY
The ability of the track to withstand applied loadings caused traffic and the environment is important if the track is to fulfil1 its purpose. Apart from vertical geometry problems which are due to voiding and deformation of the ballast and formation, the stability of .the track is also affected by lateral geometry problems caused by excessive lateral wheel/rail loadings and buckling due to high track temperatures. In the
following sections the critical lateral force to shift the track and the buckling stability of CWR will be discussed in detail.
88
1 .o
.8 . z
j
5
0 .4I
L1
0
l .
& ?
:
0
.4
2
.2
I
I
I
I
I
I
300
400
500
600
700
800
900
1000
0.2% proof stress, uy (MPa)
Figure 3.23 Variation of CJy/Oult with o;/ for a range o f AS rolled alloy railsteels (Mair and Groenhout 1978)
24722/8015
39
The critical lateral force required to shift the track
High lateral forces between the wheel flange and the rail head occur when vehicles travel around curves, and these forces are redistributed to thetrack. Amans & Sauvage (1969) have developed a semi empirical relationship to determine the critical lateral force that is required to shift the track. The reference track used in these experiments was a track that had recently been kg/m rails on resurfaced and tamped and was constructed 46 with timber sleepers at 600 mm spacings. The lateral resistanceof the reference trackI Ho(kN) was found to be H , where P
= =
2P 1 0 + 3
,
(3.88)
wheel load (kN)I applied to the track (2P load) .
=
axle
The lateral resistance of any track Ht (kN) can be expressed in terms of the reference trackI i.e.,
Ht where n t
=
nt
Ho'
(3.89)
dimensionless coefficient relating to lateral strength of track.
The lowest value of n t obtained from field measurements was T7t= 0.85, this being obtained from timber sleepered curved track at very high track temperatures.For the case of concrete sleepered track values of 'It= 1.10 were common. The critical lateral force Hc (kN) that is required to shift the track can be expressed as
=
where
r
H ~ ,
(3.90)
r
r
is a dimensionless factor
definedby
E
=
1

R poArAt ( 1 + 5 )
0.125
(E1
1 0.25
(3.91)
(F)
0
0.125 (Erx)
90
The terms p0, Ro, ko and are parameters relating to the reference track, Equation3.85; the values of which are as follows:
PO
k0
€0
= = = =
0 0.125 m 2/ C 800 m (curve radius) 20 MPa (track modulus) 0.125m0.25 0.225 N
The
other
terms
are
defined as:
cross sectional area of rail 2 ) (m temperature difference from base temperature (OC) curve radiusof track in question (m) moment of inertia of rail in the transverse direction 4 (m 1 moment of inertia of rail in vertical direction(m4) 2 Young's l.lodulus of rail steel(N/m ) track modulus of track.
CWR track
buckling
considerations
Detailed experimental research has been undertaken by Bartlett (1960) to determine the longitudinal force required to buckle CWR track. Bartlett states that the longitudinal force PB required to buckle CWR track must overcome the buckling resistances of the rail, rail fastener and the ballast, i.e.,
pB The
 rail
fastener rail ballast resistance resistance
+
+
resistance. longitudinal PB force
formula
developed
to
calculate
the
(kN) requiredto buckle the track is
91
where Is
horizontal
moment
of
inertia
of
two
4 )I rails (mm
E
‘b
S
Young’s modulus of rail steel (MPa) , length over which buckling is likely to occur(m) about 6 m,

qb Ct
sleeper spacing (m) maximum misalignment occurring over length , ( m ) say 7.5 mm, torsional coefficientof the rail fastener (kNm

rad )I and maximum lateral (kN/m).
0.5
ballast
resistance
per
sleeper
Srinivasan (1969, Chapter 9) has outlined in detail the developmen of this forumla. The expression
represents
that
part
of
the
track
panel
resistance
which
rela
to track acting as a composite strut. The expression
represents that part of the track panel resistance the torsional resistance of the rail fastener. The expression
n
relating to
represents the
that
part
of
the
track
panel
resistance
relating
to
ballast
resistance of the track.
92
A comparison
of
Bartlett's
experimental
and
theoretical
values
to buckle the track is of the longitudinal force required presented in Table 3.17. Upon closer examinationof Table 3.17, and ignoring the case of the sleepers on rollers, it is evident
that the rail resistance is of the order of 1116 per cent of the total buckling resistance, whereas the rail fastener resistance and the ballast resistance account for 1337 per cent and 5070 per cent respectively.
Unfortunately Bartlett has not indicated what the fastener system codes used in Table 3.17 represent. New Zealand Railways (Vink 1978) have carried out tests, for the range of fastener systems shown in Figure 3.24, in order to determine a representative value of the torsional coefficient, Ct, to be used with a particular fastener system. The resultsof these tests are presented in Figure 3.25. It is apparent thatof the tests conducted with timber sleepers, only new sleepers were with used, sleeper and that the variation of torsional age was not tested. fastener resistance
Sinha (1967) has assumed that the torsional resistance of dogspike rail fasteners can be taken effectively as zero. It would appear that the torsional resistance of both dogspikes and screwspikes is highly dependent upon the age and condition the timber sleepers. The factor which governs the ability of track to withstand thermal buckling is the lateral resistance of sleepers, W , which consists of three separate components: of
.
the resistance offered by the base of the sleeper (this depends upon the weight of the sleeper and the load on sleeper)
the
.
the resistance between the sides of the sleeper and the ballast in the cribs
93
7" Screwpike in Pine Sleeper
Canted
Hardwood SleeDer
5 % ' ' Screwpike
Pine Sleeper Only (Timber) Type ' P ' Bedplate
T y p e 'N' Bedplate
Assembly
Pandrol
Assembly
7" Screwpike
Pine Sleeper Only (Timber) T Y P E 'R' Canted Bedplate Assembly
Concrete Sleeper
Ribbed
P ' (concrete) Type ' Pandrol o n Concrete Sleeper
Ribbed
Canted
.
.
7" Screwpike
Pine Sleeper Only (Timber) T y p e 'R R' & Rubber Bedplate Pad
5 % ' ' Screwpike in Pine Sleeper Type ' A ' Spring Clip Assembly
Ribbed
Figure 3.24 Fastening types used i n fastener torsional resistance experiments (Vink 1978)
94
95
TABLE 3.17

COMPARISON BETWEEN THEORETICAL & EXPERIMENTAL BUCKLING RESULTS, (BARTLETT 1960)
tb
Fastening Ct (kIJm radOm5) Type
qb Nature
(mm)
of Ballast
we
(m)
Experimental Theoretical Calculations B ' ~ ' E I ~n2C (kN/m) B ' (kN) e2 3 1 (kN) 6 s q , h10 (percentage of
'%
2
I
G)
¶ h Y
11.39
4/4
9.5 8.8 8.8 7.1 9.5 6.4
38(mm) uncompacted 1.27 ballast under sleepers only 1.27 1.27 1 Ditto
710 1
1 748 60.6 27.7 11.7
1 710
899 1 1 899 919 55.8
60.2 1 28.2 748 11.6 55.2 937 28.2 16.5 27.5 1 55. 1 383 34.11 535 446 16.7 17.7 15.9 34.5 36.8
5
11.77 1.27 9.87
4 6.4 15
7.1 8.1 8.1 11.1
Ditto
1.27 513 1.27 Sleepers on rollers
1 513 50.0 1 65.5 515 515 1 030 1 030 58.5 998
13.92 10.2
W
11.1 12.7
8.25 10/3 9.67 10.1 8.25 10/3 10.1 9.62 17.0 4.13 10/3 9.8 4.81 8.25 10/3 9.62 8.25 10/3 9.62 17.8 9.1 12.7 7.7 9.1 17.8 7.1 9.8 12.7 10.2
1.27 1.27
24.2
68.8 494 31.2 60.8 1 061 15.0 27.1 1 103 14.4
20.3 uncompacted 38(mm) sleepers under ballast 20.3 only
032 1
~~~ ~~~
13.4 16.3 16.0
~ ~ ~ ~ ~~ ~~
70. 3
Ditto 17.0 12.7 Ditto Plus full boxing shoulder No Ditto Plus Full boxing and 230(mm) shoulder 1.27
~~ ~ ~~
998
~~
15.3 1 055
68 . 7 ~~
2.55 1 2.55
2.94
759
1 1 754 1 449 1 449
68.5 759 16.0 15.5
1 805 18.2 15.1 1 586
66 . 7 71. 16.5 3
12.2
2.94
1 629 69.5 18.7 11.8
.
the resistance offered by the ballast in the track shoulder to the sleeper end.
British Rail have carried out the most detailed investigation determine the value of the lateral sleeper resistance for a variety of conditions (Shenton & Powell 1973). In this investigation a summary of all previous tests was carried out and a comparison of the lateral sleeper resistance was made for a range of ballast types, ballast conditions, shoulder sizes and sleeper material types. The main results of these tests are presented in Tables 3.18 to 3.23. The results of laboratory tests presented in Tables 3.20 and 3.21 clearly show that the total lateral resistance of the sleepers depends very much the packing condition of the ballast. on
to
Field experiments have shown a large variability in the measured value of lateral resistance of sleepers. It was found on one site that there was a variation of 22 per cent along a length of 110 m, where all ballast conditions appeared to be identical. Other field measurements of the lateral resistance of concrete sleepers have shown a range of 3.5 kK/sleeper on a nominal value of 10.4 kN/sleeper. In these experiments the effects of track maintenance and consolidation operations were also quantified, and these are presented in Table 3.23 (Shenton etal. 1973)The lateral resistance of sleepers drops immediately following tamping of the track, and is gradually regained after subsequent traffic. This is clearly illustrated in Figure 3.26. It can be seen that 93 per cent of the original value of sleeper resistance is regained after 1.3 million tonnes of tr,affic; (this was equivalent to3 4 days of traffic under the test conditions).
24722/80I6
97
TABLE 3.18 Test Series 1 2
U)
 TYPICAL
VALUES OF LATERAL SLEEPERS RESISTANCE (SHENTON ET AL 1973) Resistance (kN) /sleeper Sides Base & Shoulder Sides 1.00 1.80
Sleeper and ballast Base type Wood Concrete Wood : Granite Slag As h Concrete Concrete Concrete

Total2.10 3.25 1.65 1.55
 1.30 2.70
1.00 0.85 0.85 2.60 2.00 3.17
0.90 1.10
 3.90 4.60
0.65 0.70 0.30
1.10

0.10 0.13

 0.88 1.65
~
 4.45 8.80
 2.20  2.15  1.50
3.90
W
3 4

2.52
<0.55 ~0.60 10.33
1.15
<0.20



4.11 6.15
TABLE 3.19 1Wood 2
 VALUES
~~
IN TABLE 3.18 PRESENTED AS A PERCENTAGE OF THE
~~ ~~~
TOTAL
~ ~~~~~
RESISTANCE
LOO 100
100 100 100
Wood Concrete 30
:
29
 47 3058
45 40 55
65
37

65 54

5 4
30
~
25
28
Granite Slag As h Concrete Concrete 3
30 30 20 30
60
25 25 5 38
100
100
100
50 Concrete
50

4


It can be seen that ingeneral the base resistance of the sleeperis thc grcatcst proportion, this being approximately 50 per cent of the total resistance. Thc resistance provided by the sides of the sleeper is of the orderof 20 to 30 pcr cent of the total.
TABLE 3.20
 TEST
Slze
SERIES 1, EFFECT OF SHOULDER SIZEUPON THE INCREASE OF LATERAL SLEEPER RESISTANCE, (SHENTON ET 1973) AL
kN/sleeper Resistance Ballast Sleepers State Only
3 8 mm 3 8 mm 38 mm 3 8 mm
Only Wood Concrete Concrete
P
0 0
150 mm Shoulder 230 mm Shoulder 300 mm Shoulder TotalShoulderTotalShoulderTotalShoulder Only
Loose 2.12 Compact 0.74 3.45 Loose Compact
3.41
0.10
2.46 3.52 0.46 3.72 5.06 3.24 8.50
0.44 0.88 3.59 0.80 0.51 1.76 0.13 1.30
0.69 2.71 4.39 6.31 3.48 8.80 1.13
0.15
75 mm Loose 75 mm Compact
0.10 1.65
These values are all slightly higher than the results from series 2. A few results available from test series 5 can be summarised as: size Shoulder
305
(mm)
Resistance kN/sleeper
6.46 6.82 6.79
460 610
These
show
that
no
benefit
is
gained
by
increasing
the
shoulder
Size
beyond 460
m m .
TABLE 3.21

TEST
SERIES 1, COMPARISON OF TIMBER
AND
CONCRETE BALLAST
SLEEPERS LATERAL RESISTAXCE FOX VARIOUS CONDITIONS (SHENTON ET AL 1973) ~
~ ~
______~
Resistance kN/sleeper Sleeper
3 8 mm 76 ~ilip. LooseCompactLooseCompact
Component of Resistance Ballast
Wood
1.01
1.01
1.03 1.83
0.90
1.28 Base Friction
~ ~~~
Concrete
Wood
2.14
1.01
2.68 1.70 1.87
2.64
3.90
Side Concrete
Wood
Friction
1.11
0.80 0.44 0.46 3.51 2.46 5.06 3.71
1.29 0.30 0.28
4.56 2 2 9 m Shoulder
Concrete
Wood
0.51
1.30 0.13 2.21 3.25 8.
Concrete
4'44 5o
Total Resistance
TABLE 3.22
 TEST
SERIES 2, COMPARISON OF TIMBER AND CONCRETE LATERAL SLEEPER XESISTANCE ON GRANITE, (SHENTON
ET AL 1973)
S leeper
Resistance kN/sleeper Base Sides
3 O m m Shoulder Total
Soft Wood Concrete 1.10 0.25
1.00 0.40
0.65
2.60
2.05 3.95
It would appear resistance
that from
the the
greatest
part of the
increase
in
total
comes
increase in weight of the sleeper.
101
TABLE 3.23
 EFFECT OF
VARIOUS MAINTENANCE OPERATIONS ON THE MAGNITUDE OF THE LATERAL RESISTANCE OF CONCRETE SLEEPERS (SHENTONET AL 1 9 7 3 )
% Loss Lateral in Resistance
Maintenance Operation Tamping only Tamping and shoulder
16
 37
12
consolidation
Tamping and crib and shoulder consolidation 15
Having calculated the longitudinal force required to buckle the track pBr the track temperature over and above the stress free temperature t , ( deduced by (3.93)
0
C)
atwhich
this
buckling
will
occur
can
be
E 2Ar a t
*
longitudinal force required to buckle the track (kN) 2)* cross sectional area of two rails (mm coefficient of thermal expansion (OC') I and
~
Young's Modulus of rail steel (MPa).
For curved track Magee (1965) has developed a simple empirical equation to determine the lateral forces per unit length of the rail Pf produced by continuously welded rail subject to a change in temperature. This equation expressed in metric units is Pf where Pf At

At 86.3
(3.94)
=
= =
D ,
total latera1,forceper metre of track length (kN/m), temperature difference from base temperature ( O C ) , degree of track curvature (degrees).
102
Before tamping
After
tamping
l
l
1
l
I
!
0.0
0.2
0.4
0.6
0.8
1 .o
l 1.2
1.4
Cumulative Volume of Traffic (Mtgross)
Figure 3.26 Relationship between sleeper lateral resistance and cumulative volume of traffic, showing the influenceof the tamping operation (Shenton and Powell 1973)
103
This
equation
was
originally
developed
for
continuously
welded
rail track elevated on timber trestles. Prause et al. (1974) state that if reliabile data are available for the lateral
resistance of sleepers, for specified ballast types and cross in curved sectional geometry, the maximum sleeper spacing S max track to prevent lateral buckling (from simple stability conside ations) is given by
' m a x where
RL
= =
I
(3.95)
lateral resistance of sleeper to movement in the ballast (kN). conventional design of railway track relating to
Details of the
the analysis of sleepers (i.e. rail seat load determination, contact area and flexural strength assumptions) are presented in Chapter 4. RAIL WEAR LIFE The equations that have been developed for rail wear life
pre
diction are all based upon empirical measurements and have real theoretical backing. In the following section three major
no
methods of estimating rail wear; the University of Illinois method, the AREA method and Couard's method, will be presented
detail. An Australian formulaused in rail wear life prediction, the Westrail formula, will also be discussed. The main criticism of the Universityof Illinois and the AREA methods for determini the life of rails is that there are no guides for improving their wear performance (Prause et al. 1974). It is also not apparent whether the influence of other conditions that effect the life rails, such as rail breakages, rail shelling, transverse defects etc., are included in the following methods. The effect of transposing the rails resulting an in increase in rail wear life is also not treated explicitly in any of the following methods.
104
The
University
of
Illinois
Rail
Wear
Formula
Field measurementsof rail wear have been carried out afor number of U.S. railway systemsby the University of Illinois
to
determine the service life of continuously welded rail (Hay et al. 1973, 1975). Rail head wear data were obtained in the field from the Burlington Northern System (BN), Norfolk and Western Railway (N&W), Illinois Central Railroad (IC) and the Atchinson, Topeka and Santa Fe Railroad (AT&SF). Observations were recorded by the and using a planimeter theloss of rail head area caused passage of traffic during a specified time period was measured. With the advent of continuous welded rail which has virtually eliminated rail joints, the primary causes of rail head wear can be considered to be due to abrasion between the wheel and the rail, plastic flow of the rail head to due excessive wheel loads or, a combination of both. Abrasion is influenced in turn, by such factors as curvature, traffic, speed, type of rolling equipment maintenance standards and the operating environment.
.
of Illinois (Hay et a1 The formula developed by the University 1973) to calculate the anticipated abrasive rail head area wear caused by traffichas, in order to determine the annual rail head area wear, been restated as
(3.96)
where
W , Wt
= 
annual abrasive rail head area wear (mm2/year), average rail head area wear term, defined a for 2 /Mt particular rail size and for tangent track (m gross) , degree of curve (degrees), annual gross tonnage(Mt gross/year), and wear factor varying with the degree of curve, i.e. the amountby which the tangent wear value must be increased to equal the measured curve wear value.
24722/60"17
105
The average rail head wear from field data for tangent track commonly used rail sizes are listed in Tables 3.24 and 3.25. The average head wear values are weighted to account for the volum of traffic over each test section. It should be noted that the method of weighting (outlined in the footnote to Table 3.24) assumes a linear relationship between head wear and traffic volume in order to relate different railway systems.
Some errors and inconsistencies of data presentation were discovered in the original source reports and these have been ified.
re
It should be noted that the field data presented in Tables 3.24 and 3.25 define the average head area wear in terms of a 100 million gross tonsof traffic carried. Here tons refers to U.S. short tons and the conversion to equivalent metric units are presented at the bottom of each table. The value of wear factor, K , for any particular degree of curvature of the track, was
calculated by comparing the average rail head wear obtained in curved track with the average rail head wear obtained in tang track and defining the relationship as Average rail wear in curved track Calculated valuesof Kw are listed for
=
Wt (1 + Kw DC). various degrees of track
curvature and rail size in Tables 3.26 and 3.27. It should be noted that for 57 (kg/m) rail, field data were only available for curves of0°15' through to 3 ' 0 ' whereas for'66 kg/m rail field
data were only available for low degree curves et (Hay al. 1973). Therefore the assumed and extrapolated values given in Tables 3.26 and 3.27 should be treated with reservation until more rail wear data for the higher degree curves becomes available. Specific formulae for determining the amount rail of head area wear, derived from the results of Tables 3.26 and 3.27 and written in the form of Equation 3.96 are as follows:
106
TABLE 3.24
 SUPWRY OF
THE AVERAGE FIELD TANGENT TRACK RAIL VARIOUS U. S. RAILWAY
mm
DATA FOR 5 (kg/m) 7 RAILSIN SYSTEMS, (HAY ET AL 1973)
Railroad Average Head Wear Average Weight (a) Weighted Average per 100 YGT Tonnage Factor Head Wear (a) 2 2 (MGT) 2 2 (in 1 (mm 1 (in ) (mm ) BN 62.55 43.29 0.0671 N &W 0.0850 54.84 180.90 0.2457 2.89 158.52 IC 41.16 0.0638 1.54 96.51 26.71 0.0414 0.0472 AT&SF 0.2237 296.35 30.45 144.32 4.74
1.00
43.29 0.0671
10.17 0.6003 387.29
(a) Lowest average tonnage (62.55 for the BN) taken as base (1.00) and divided into the other tonnages to obtain a weighting factor. (eg. 180.9 I 62.55 = 2.9, for the N&W values). NOTE: Average head wear of 57 kg/m rail for tangent track is 2 387*29  38.08 mm /l00 MGT “i7= 0.420 mm 2/Mt gross.
TABLE 3.25

SUMMARY OF THE AVERAGE FIELD TANGEfJT TRACK RAIL DATA FOR 66 (kg/m) RAILS IN VARIOUS U.S. RAILWAY SYSTEMS, (HAY ET AL1973)
WEAR
Railroad Average Head Wear Average Weight Weighted Average per 100 MGT Tonnage Factor Head Wear 2 2 (mm2) (MGT)(a) (in ) (in )
N &W
2 (mm )
IC
SF
0.0720 46.45 94.8 2.59 245.67 24.97 0.0387 2.27 215.51 47.87 0.0742
46.45 0.0720 1.00 64.65 0.1002 108.64 0.1684 5.86 219.74 0.3406
NOTE: Average head wear of 66 kg/m rail for tangent track is
(a) U.S.
Short tons. 107
(b) Metric tonnes.
TABLE 3.26
 SUGGESTED K"
ET AL
VALUES FOR USE WITH 57(kg/m) RAIL, (HAY (b) 1973)
El
Rail
High 0.20
(a) Curvature
Low Rail
0.10
0.40 0.60 0.80
1.00
 0°59' 1OOO'  1O59' 2OOO'  2O59' 3OOO'  3O59' 4OOO'  4O59' 5OOO'  5O59' 6OOO'  6O59' 7OOO'  7O59' 8OOO'  8O59'
ooO1'
0.20 0.30
1.00 1.00
1.00
1.00
goOO ' and
over
1.00
0.40 0.50 0.60 0.70 0.80 0.90 1.00
(a) Degrees of curvature are based upon the American practice of using 100 ft chord lengths.
(b)
K value based on groupings around low degree curves from field data and interpolated therefromK to = 1 for 8degree curves. RAIL, (HAY LUES FOR USE WITH 66 (kg/m)
ET AL Curvature (a) Rail
1973) Low Rail High 0.46 0.50 0.54 0.58
0.62
ooO1' 2O29' 2O30' 2O59' 0.15 3OOO' 3O29' 3O30' 3O59' 4OOO' 4O59' 5OOO'  5O59' 0.37 6OOO' 6O59' 0.46 7OOO' 7O59' 0.54 8 ' 0 0 ' and over


0.11
0.20

0.22
0.28
0.71 0.78 0.87 1.00
0.63
(a) Degrees of curvature are based upon the American practice of using 100 ft chord lengths.
(b) K value based on groupings around low degree curves from field data and interpolated therefrom K to = 1 for 8degree curves.
10 8
(a) For 57 (kg/m) plain carbon CWR:
W ,
=
0.420 (1 + Kw DC) DA
(3.97)
(b) For 66 (kg/m) plain carbon CWR:
(3.98)
where
W , K ,
D ,
= =
=
=
2 annual rail head area wear (mm /year) I wear factor varying with degree of curve, (values presented in Tables3.26 and 3.27), degree of curve (degrees) , and annual gross tonnage(Ilt gross/year) .
2 Denoting the permissible limit of rail head wear by @A (mm ), the expected lifeof the rail T (years) can be defined by
Y
T
Y
@A = 
wa
(3.99)
Consequently, the expected life of rail T (Mt gross) can therefore be definedby
T
=
Ty DA
.
(3.100)
The AREA Rail Wear
Formula
Underlying the following analyses is the basic assumption that the track structure behaviour can be established fora set of "base" track/traffic conditions and that variations from these base conditions can be reflected a by process of factor analysis, i.e. that basic track structure performance can be "factored" to predict relative effects of differing track/traffic environments (Danziq, Hay & Reinschmidt, 1976). The method developed by the AREA (based upon available research data developed by the AREA and by several individual railroads) evaluates rail life as a function of the following track and traffic characteristics:
109
annual tonnage density(Mt gross) composite of rail type, weight, rail joint type, metallurgy and condition actual average train speeds for a particular train service type and traffic axle load class gradient of the track curvature of the track and the use of lubricants in the t actual traffic axle loads for,a particular train services type and train speed class.
For tangent, level, fishplated rail track under mainline speed conditions, here assumed to be 80 (km/h), the life of a rail can be estimated by T where
T
DA
=
1.839 KT RWt
0.565 (1.102 DA) I
(3.101)
= 
%t= KT

rail life (Mt gross), annual gross tonnage (Mt gross/year) , rail weight (kg/m), and composite constant reflecting track the level of maintenance.
conditions
and
This formula was derived using imperial units, therefore the value 1.102 represents the conversion of tonnes U.S. to short tons, while the value 1.839 represents the product of the conversion ofkg/m to lb/yd with the conversion of U.S. short tons back to tonnes. For the case of tangent, level, fish plated plain rail carbon track, the value of KT is generally adopted as being 0.545. This value was developed from surveys U.S. of industry experience. A more general formula which adjusts the rail life estimate tangent track to allow for a wide range of track and traffic conditions has been expressed as (Danzig et al.1976)
i
110
1.565 1.839 KC KG KR RWt (1.102 DAj T = (3.102)
where
T
= =
RWt= KC 
n
=
rail life (mt gross) I annual gross tonnage (Xt qross/year), rail weight(kg/m) , track curvature and lubrication factor (preliminary values are presented in Table 3.28), track gradient factor (preiirninary values developed , for tangenttrack are presented in Table 3.29) rail factor (which includes rail weight, rail strength and joint type; preliminary values are presented in Table 3.30) , total number of differext typesof traffic in the overall annual gross tonnage, subtonnage (Mt gross/year) , which relatesto a particular speed class, axle load class and service type I speed class factor (that relates to a particular subtonnage Di (preliminary values are presented in Table 3.31) , wheel load class factor, that relates to a particular subtonnage Di (preliminary values are presentee in
KS
i

Table 3.32) , and service type factor, this factor relates to the type of traffic that is represected by the subtonnage Di. (Preliminary values of R have been suggested as;
0.91 for UnitTrain O2eraEions and 1.0 for other traffic) .
The conversion factors in Equation 3.L02 are thoseas previously mentioned in Equation3.101. It shorrld be recognised that the majority of the reduction factors listed i r . Tables 3.28 to 3.32 are preliminary; some being entirely hased u20n juiigeir.ent, and have as yet not been based upon ad'equate experimental data.
Prause et al. (1974) expresses Equation 3.102 in a different form to enable the prediction of rail life for tangent track in years to be determined, i.e., 111
TABLE 3.2 8
 RAIL
(a)
LIFE ''KC1' FACTOR
 CURVATURE
AND
USE
OF
CURVE
OILERS, (DANZIG ET AL 1976) Curve Factor Class (degrees) Factor (oilers)
1.00 1.00
 0.5  1.5 1.5  2.5 0.88 2.5  3.5 0.79  4.5 3.5 0.70 4.50.62  5.5 5.5  6.5 0.55 6.5 0.48  7.5 7.5  8.5 8.5  9.5 0.40
0 0.5
0.87 0.74 0.61 0.49 0.38 0.30 0.22
0.16
1.00
0.44 0.37
' 9 . 5
0.12 0.10
(a) Degrees of curvature are based upon the American practice of using 100 ft. chord lengths.
TABLE 3.29
 RAIL
1976)
LIFE "K,"
U
FACTOR
 TRACK
GRADIENT, (DANZIG ET AL
Gradient Class
Factor grades grades grades
1.0
 0.5 per cent 0.5  1.0 per cent 1.0  1.5 per cent 1.5  2.0 per cent
0
>2.0 per
grades cent grades
1.0 0.9 0.8 0.7
112
TABLE 3.30
 RAIL
1976
LIFE "Y,"FACTOR
 RAIL WEIGHT(a)
Rail <55 kg/m 5561
(DANZIG ET
AL
~
~~
Rail, Type Metallurgy, Condition
Weight
kg/m
z 61 kg/m
Fishplated Plain (Control Cooled) New Fishplated Plain (Control Cooled)
S.H.
0.545 0.545 0.545 0.545 0.545 0.545 0.774 0.964 n/a
S .H.
CWR, Plain (Control Cooled) New CWR, Plain (Control Cooled) 23.8m, Plain (Control Cooled) 23.8m, Plain (Control Cooled) FlameHardened 23.8m, New FlameHardened 21.0m, S.H. 23.8m,Silicon High New New
S.H.
n/a 0.774 0.964 0.545 0.545 0.545 0.545 0.545 0.545 n/a 0.545 0.545 0.545 0.545 E/ a n/ a
0.812
0.567
(a) These preliminary values were based upon rail research conducted at University of Illinois (Hayet a1 1973) and are recommended by the AREA.
24722/8018
113
TABLE 3.31
 RAIL
LIFE "KV'' FACTOR
 TRAIN
SPEED (DANZIG ET AL
1976) (a)
Speed Range (km/h) 108.6 100.6 92.5 84.5 76.4 68.4 60.4 52.3 44.3 36.2 28.2 20.1 12.1 Assumed Speed (km/h) 112.7 104.6 96.6 88.5 80.5 72.4 64.4 56.3 48.3 40.2 32.2 24.1 16.1
K
 Factor
0.80 0.85 0.90 0.95
1.00
 116.7  108.6  100.6  92.5


84.5 76.4 68.4 60.4 52.3 44.3 36.2 28.2 20.1
1.05
1.10
1.15 1.20 1.25 1.30 1.35 1.40
(a) Based upon workof Prof. Talbot (AREA),this work predicted increases in dynamic impact on the track structure of 1 per cent per mph increase in vehicle speed for speeds in excess of 5 mph (Talbot 19181934).
TABLE 3.32
 RAIL
1976)
LIFE

"K" FACTOR
 WHEEL
LOAD, (DANZIG ET AL
e
Load
Wheel
67 116 133 147 156
 116  133  147  156  167
> 167
67
1.153 0.862 0.841 0.545 0.439 0.381 0.325
114
(3.103)
where
TY
= =
rail life for tangent track (years) I annual tonnage density (Mt gross/year), rail life (Mt gross), and permissible limit of vertical head wear (m). wear
T
=
=
ev
and 4.76 mmrepresents the commonly assumed vertical head used as the limit of rail limit of 3/16 inch, and as such was life inEquation 3.102. The Couard Method of Calculating Rail Wear
The Couard method for calculating the annual rail wear based on the following equations (Prause et al. 1974): (a) Vertical Rail Head Wear on Tangent Track: W ,
rate
is
=
8.70.10 5 Q1 V DA (1 + 0.23 Gls7) + 0.0635.
(3.104)
(b) Vertical Rail Head Wear on the High Rail of Curved Track:
W ,
=
9 . 5 7 . 1 0 ~Q, ~
v
D~ (1 + + 0.23 G1'7) + 0.0635.(3.105) 254
(c) Side Head Wear on High Rail of Curved Track: W ,
=
1.39.10
4
Q 2 V DA DC (1 + ub 25 4 + 0.23 G l S 7 ) + 0.0635.

(3.106) (d) Side Head Wear on Curves with Rail Lubricators:
(3.107)
where
W , W ,
=
= =
annual vertical rail head wear rate (mm/year), annual side rail head wear rate (nm/year), annual side rail head wear rate for lubricated rail (mm/year)I
115
G
V
D ,
Q~
= = = =
track gradient (per cent), annual traffic density (Mt gross/year), operating speed (km/h) , unbalanced superelevation (mm) , degree of curve (degrees) (1), ratio of rail head width to that of the 69.5 rail, and
ub =
=
=
(kg/m),
Q2
ratio of rail head depth to that of the 69.5 kg/m, rail.
The constant 0.0635 mm/year in the above equations represents the assumed rate of annual headloss attributed to the effects of rail corrosion. wear
The Couard method for determining the annual rail head was originally developed for a track with 69.5 kg/m, (140 lb/yd RE), rail. Therefore, the rail size correction factors, Q1 and Q*, have to be applied to the equations when smaller rail
sizes are used. Values of the rail size correction factors for Australian and American rails are presented in Table 3.33. It should be noted that the coefficient expressed in Equations 3.104, 3.105 and 3.106 also contain the conversion U.S. of short tons to tonnes.
By dividing the annual wear rate W , or Ws into the corresponding permissible limit of rail head wear the rail life in years predicted. For the case of curved track the rail life is the less of
c
TY where TY
"
orT
wV
Y
= 0 s , wS
(3.108)
ev os
= =
=
WS
rail life (years), permissible limit of vertical rail head wear (mm), permissible limit of side rail head wear (mm), and are as above.
(1) Establish a maximum allowable wear (mm/year).
116
TABLE
3.33
 WEAR
FACTORS
FOR
CALCULATING
RAIL
HEAD BY WEAR THE
COUARD METHOD (PRAUSE ET AL 1974) Rail
(kg/m)
Section
Q1
Wear
Factor Q2
31 41 47 50 50 RE 53 57 RE (a) 1.25
1.20 1.20 1.09
1.09
1.21
1.17
59 (a) 60 66 68 69 RE
1.12 1.09 1.10 1.13 1.09 1.09 1.02 1.00
1.22 1.10
1.06 1.00
(a) New Australian rail section.
The life of the railT (Mt gross) can be defined by Equation 3.97. Several significant factors which affect the rate of rail wear
are not consideredat present by the Couard method. These factors include the rail steel composition, the track environment (dry or wet, corrosive and/or abrasive), and the extent of lubrication. Prause et al. (1974) consider that the currently available methods for estimating the rail wear on curves are not entirely satisfactory. However, an improved procedure could be developed using an empirical relation similar to the Couard formula and modifying the results with corrective factors. This empirical relation should be verified and modified and should also be based on data from wear test metallurgical experiments using typical wheel and rail steels. The following procedure was suggested:
117
(1) Establish a maximum allowable wear (mm/year).
( 2 ) Establish a base wear value for a given traffic density, (Mt
gross/year) moving at a given rail steel (=O .5 per cent C).
average
velocity
for a standard
(3) Modify these results for the degree of curvature.
(4)
Multiply by an environmental factor.
(5) If the predicted wear is too high, estimate the amount of
wear reduction that could be obtained by alloy using the relation that wear is reduced about 8 per cent for every 0.1 per cent increase in carbon or carbon equivalent.
(6)
Apply a factor for lubrication if further reduction in wear is desired or if alloy steel is undesirable. Wear life can be increased Rail considerably Life by lubrication.
The
Westrail
Formula
A formula which predicts the expected life of rails in track has been developedby the Western Australian Government Railways (Westrail 1976, Hoare and Payne 1978). This simple formula expresses the life of the rail as a function onlyof its weight, and incorporates AA rail life data for rails with weights abov 50 kg/m (AREA, 1951), with Westrail rail life data for rails with weights below30 kg/m, see Figure 3.27. developed for jointed rail track and is T where T R W t ' 35.6. 106 RWt 3.98
r
The formula was stated as
(3.109)
=
rail life (Mt gross), and rail weight (kg/m).
118
80
70
60
50
E .
I
Y
m
.
I
T = 35.6X 106 R , , 3 . 9 8 /
2 35
(OI
Rail Life(Mtgross) Legend:
@ Pennsylvania Railroad @ Illinois Central
0
W.A.G.R.
T = Life of Rail ( M t gross)
R W t = Weight of Rail (kg/m)
Figure 3.27 Approximate relationship between rail weight and expected r a i ll i f e (Westrail 1976)
119
The formula does not give an indication of what annual tonnage, or what axle loading carried is inherent in the relationship. It is also not apparent whether this formula was developed for tangent or curved track, or both; and whether it incorporates other conditions that influence the life of the rail, such as rail breakages, rail shelling, transverse defects etc. Permissible limits of rail head wear
The permissible limits of rail head wear can be defined in the following interrelated geometric ways:
. .
alloweable vertical head wear loss allowable side head wear loss allowable head area loss maximum angle of side wear.
. .
The definition of these permissible limits at can, best, be regarded as subjective, the values of which largely depend upon the railway operator. In the following discussiona comparison of typical rail head wear limits is made.
Although varying with rail section, an average rail head vert wear of 4.76 mm, (3/16 in.) is sometimes taken as limiting the U.S. mainline track (Prause et al. 1974). useful life of rail in , It should be noted that the rails may be removed from the track for reasons other than abrasive wear of the rail head above. Such factors as the occurrence of an excessive number of rail breakages, or fatigue defects, per kilometre will also signific influence the actual life of the rail. Rails are not normally worn to the condemning limit by imposed
the bending stress criteria. Before that point is reached, the wheel flanges strike the upper edges of the fish plates in th
case of jointed rail track. Although the bar may then wear alon with the rail, good practice warrants removal of the rail at stage (Prause et al. 1974).
120
, ,
For CWR the
vertical
wear
criteria of when
the
wheel
strikes
the
fishplate may be excessively restrictive, as much of this type of joints. The long of track is constructed with small numbers string lengths from the flash butt welding plant are thermit welded in the field into lengths of several kilometres. When strings of CWR are secured with fish plates, such as at insulated joints, the relatively few joints per kilometre would pose .no excessive risk in allowing the wheel flange to make contact with the fish plate. Under those conditions, the fish plate will wear downward as the rail head wears downward. An unfavourable condition could exist, however, a if worn fish plate had to be replaced with one not worn. Milled fish plates could be provided for such situations. By the time the rail had worn to this degree, it would undoubtedly have been placed in low speed track (Hay etal. 1973). For these reasons, Hay has used a fish plate clearance of 0.8 mm (1/32 in.) as thecondemning limit for the 57 and 66 kg/m rail sections analysed. This represents an equivalent head wear area of 350.3 mm2 for the 57 kg/m and 518.7 mm2 for the 66 kg/m rail (Table 3.34) . These head wear areas correspondto 14.0 and 18.5 percent loss respectively in the total head area of the two rail sections. The AAR have established permissible limits for rail head wear based upon the operating train speeds over a particular class of track (Railway Track and Structures 1970). These limits expressed as percentages of total head area and probably defined CWR fortrack are presented in Table 3.35. An envelope of allowable rail wear limits, incorporating bothloss of rail height and side head wear + due D flange contact, can be drawn to suit the conditions of particular railway operators; for example those based upon Canadian National recommendations €or various rail section are presented in Figure 3.28 (King 1976). These limits correspond quite closely with those recommended by Hay et al. (1973), Table 3.34. All railway systems have established similar rail head wear condemning limits.
24722/8019
121
TABLE 3.34
 LIMITS
OF RAIL HEAD WEAR, (HAY ET AL 1973)
Rai 1 Criteria Condemning Weight Bending Stress Criteria Fish Plate Clearance(a) (kg/m) Factor of Vertical Head Equivalent Area Vertical Head Equivalent Area Loss of Wear with Safety of Wear Wear of Wear Rail Head Area 2 2 2 2 One S leeper (mm 1 (in) cent) (mm) (in 1 (in) (mm) (in 1 Missing (mm ) (per 5714.0 350.3 0.543 6.351/4 1.11 438.7 0.68 7.94 5/16 66 1.93 19.05 3/4 1.34

1 18.5 518.7 0.804 8.73 11/32 245.2
(a) Recommended by Hay.
P
W
TABLE 3.35 Class
 AAR LIMITS
FOR
RAIL
HEAD WEAR (RAILWAY TRACK & STRUCTURE
1970)
of
Rail
Train Speeds Maximum Permissible Reduction in Rail Head Area (per cent), for Various Rail Sections (kg/m) 354 2 4249 4956 5664 >64 (km/h) <35
100
 130 75  100 50  75 25  50 24 20 0  25 28 24
20
30 32 34 36 36
30 32 34 40 38 42
34 38 42 46
36 38 44 50
8
7
1
M a x i m u m loss of height = 6.35 (mm) a l l sections. (Based u p o n flange contact with fish plates)
I
6
I I
5
I
i g
'5 I
LC
v)
E
E
4
l
g
3
2
1
a
1
4 2
3
Side Head
5 Wear (mm)
6
7
8
Note: Values used in drawing the envelope of rail wear limits are:
LossHeight of
(mm) 66 (kgirn) 7.92 4.75
75
6.35
1.57 3.18 4.75 3.18
Side Head Wear 57 (kg/m) 3.18 6.35
(mm) 50 (kg/rn)
G42 h i m )
4.75 1.57
1.57
Figure 3.28 Envelope of rail wear l i m i t s for loss o fr a i l height and side head wear caused by flange contact; Canadian l i m i t s (King 1976)
12 3
It is
notable
that
the
permissible
limits on the
amount
of
rail
head wear do not adequately relfect rail strength levels, nor they relatedto vehicle dynamic tracking, which apart from other factors, is influenced by changes in the track gauge and the rail cant. Rail replacement the constitutes rail head a major track maintenance limits, for cost,
consequently
wear
condemning
together
with
the optimum worn rail head condition should be given more detailed study.
suitable
transposing
12 4
CHAPTER 4
 SLEEPER ANALYSIS
The functions of the sleepers are to transfer the vertical, to the ballastand lateral and longitudinal rail seat loads formation, and to maintain the track gauge and alignment by providing a reliabile support for the rail fasteners. The vertical dependent loads upon subject the the sleeper to a bending of the moment which is the condition ballast underneath
sleeper. The performance ofa sleeper to withstand lateral and longitudinal loadirlg is dependent upon the sleeper's size, shape, surface geometry, weight and spacing.
Before the sleeper can be analysed in terms of its capacity to withstand the bending stresses caused by the vertical rail seat loads the sleeper support condition and its effect upon the contact pressure distribution must be quantified. The contact pressure distribution between the sleeper and the ballast is mainly dependent upon the degree of voiding in the ballast under the sleeper. This voiding is caused by traffic loading and is due to the gradual change in the structure of the ballast and the subgrade. The exact contact pressure distribution between the sleeper and the ballast and its variation with time will be of importance in the structural design of sleepers. It is practically impossible to predict the exact distribution for a sleeper in the intrack condition (ORE 1969). In order to calculate the sleeper bending stresses a uniform contact pressure distribution between the sleeper and the ballast is assumed to occur in practice; thereby enabling a visual interpretationof how the vertical force exerted by the rail on the sleeper is transmitted to the ballast. Various hypothetical contact pressure distributions between the sleeper and the ballast and their corresponding sleeper bending moment diagrams as postulated by Talbot (19181934) are presented in Figure 4.1.
125
Distribution of bearing pressure Uniform pressure
Flexure of sleeper produces variations from (a)
Moment
d===&zJ
r
5 !
9
Tampe side d either Laboratory test
of rail
1
Principal bearing on rails M a x i m u m intensity at ends M a x i m u m intensity in middle Centrebound Endbound Welltamped
L
It
1 x 
i
A
Depression at railseat
z
*
Figure 4 . 1 Hypothetical distribution of sleeper bearing pressure and corresponding sleeper bending m o m e n t diagrams for a sleeper (Talbot 1920)
126
When the track is freshly tamped the contact area between the sleeper and the ballast occurs below each rail seat. After the track has been in service the contact pressure distribution between the sleeper and the ballast tends towards a uniform pressure distribution. This condition is associated with a gap between the sleeper and the ballast surface below rail the seat for timber sleepers (Talbot19181934). The condition of centrebinding of timber and concrete sleepers tends to develop when maintenance is neglected. Hence it can be readily seen that the contact pressure distribution between the sleeper and the ballast is a time dependent, i.e. cumulative traffic tonnage, variable. In order to prevent centrebinding of the sleepers the ballast is tamped under therail seat. The current maintenance of concrete sleepers usually requires the provision of a nonpressure bearing centre section of the sleeper. This width is arbitrary, but is usually about 500 m (ORE 1969). The provision of this centre gap is to ensure that the sleeperwill not become centre bound, thereby preventing detrimental bending stresses from occurring. Referring to the flowchart of the conventional railway track design procedure, presented in Figure 1.1, it is apparent that the selection of the required sleeper size and spacing is determined by an iterative trial and error solution. A sleeper size is adopted anda sleeper spacingis selected on the basis of being the maximum spacing which satisfies the sleeper design constraints. These being principally the contact pressure between the sleeper and the ballast and the sleeper flexural capacity. If the sleeper size chosen is unable to satisfy these constraints, or if the sleeper spacing thus determined is considered inadequate, the procedure is repeated for another sleeper size. Details of timber railway systems sleeper are dimensions in currently used by Australian presented Table 4.1.
Before the contact pressure between the sleeper and the ballast can be calculated the following factors be must quantified:
127
TABLE 4.1 Railway
 TIMBER
System
SLEEPER
DIMENSIONS
AND
SPACINGS
CURRENTLY
USED BY AUSTRALASIAN
RAILWAY
SYSTEMS
(GORDON 1973)
Gauge
(mm)
(ft.in) 4'84" 4'84" 5'3" 4'8%''
Dimensions Sleeper Sleeper (length X breadth X thickness)
(mm)
New South Wales Commission Transport Victorian Railways Australian
(a)
(ft.in) 8'0"x 9"x44" 1641 610 610
No per km
Spacing (mm) No per (in) mile 2640 2640 625 2580 24 24 24
Public 1435 1435 1600 1435 Railways 1067 1600 1600 1435 4
2438x229~114 2440x230~130 2743x254~127 9 2591x254~127
1641
1600
'O"xlO"x5" 8' 6"xlO"x5"
National
Commonwealth Railways 1435
4 '8tf" 3'6 5' 3" 5'3" '8%" 4'84" 3'6" 3'6" 3'6"
2438x229~114 1981x203~114 2591x254~127 2591x229~165 (a) 2591x254~127
8' 0"x 9"x4%" 6'6"x 8"x44"
1641 1312
610 760
2640 2112 2420
24 30 26
8'6"1O"x5" 665 1502 8'6"x 9"x6%" (a) 8'6"xlO"x5"
p N
(b)
South Australian Railways 1435
1067 1067
m
2591~229x165(~) 8'6"x 9"x64" (a) 1981x203~114 6' 6"x 8"x44" 1981x203~140(~) 6'6"x 8 " ~ 5 4 " ( ~ ) 2134x229~127 2438x229~114 8 2134x229~114 2134x229~114 2134x229~152 2134x263~127 2134x263~152 2134x229~114 2591x229~152 2667x229~152 2591x229~152 7'0"x 9"x5" 1641
610
( c ) Tasmanian Railways 1067
~ ~~~ ~ ~~ ~ ~~ ~~ ~ ~ ~
2640 2640 2500
24 24 25.3 24
Western Australian Government 1435 1067Railways Queensland Railways New Zealand Railways 1067 1067 1067 1067 1067 1435
4 ' 8 : " 3'6
3'6" 3'6"
3'6" 3'6" 3'6"
'0"x 9"x44" 1641 610 7'0"x 9"x44"
7'9"x4%" 0"x 1551 645
7'0"x 9"x6"
7' 0"x 8"x5" 7'0"x 8"x6" 7' 0"x 9"X44"
8' 6"x 9"x6"
1641
610
2640
~~
Mount Newman Mining C O Hamersley Iron Railway 1435
~~~ ~~ ~~ ~~ ~~ ~ ~ ~~ ~~ ~ ~
4'84" 4 '8%"
1876 2020 1790
530 495
560
3020 3250 2880
21 19.5 22
8'9"x 9"x6"
~
Cliffs Robe River Mining
CO
1435 4
'84"
8'6"x 9"x6"
(a)
Denotes softwood sleepers (all others specified are hardwood)
~
.
the effective sleeper support area beneath the rail seat the maximum rail seat load occurring at a sleeper. factors are discussed in detail in the following sections.
.
These THE
EFFECTIVE
SLEEPER SUPPORT AREA BEXEATH THE RAIL SEAT
The effective sleeper support area beneath rail the seat is defined as the product of the breadth of the sleeper and the assumed value of the effective length of sleeper support at the rail seat. The AREA defines the effective length of a sleeper as being the distance from the end of the sleeper to the point inside of the edge of the rail base over which tamping operations extend. Magee (19651971) assumes that this distance is 300 mm for 2440 X 200 X 150 mm sleepers,380 mm for 2590 X 230 X 180 mm sleepers and 466 mm for 2740 X 250 X 200 (mm) sleepers. Resulting of approximate1.y 840, 990 in assumed sleeper effective lengths and 1140 mm respectively. The effective length of a sleeper can also be deduced by using either of the relationships developed by Clarke (1957) or by Schramm (1961). Clarke defines the effective length of sleeper support L (mm) under the rail seat for timber sleepers as
L
where
R g
= = = =
(.eg)
[1
(4.1)
125 t
t For
total sleeper length (mm) distance between the centreline of the rail seats (mm) and sleeper thickness (mm). sleeper lengths this equation can be approximated by
common
24722/8020
12 9
There are no comparable equations derived for concrete sleepers in which the effective length of sleeper support incorporates the effectof sleeper thickness.
Schramm defines the effective length of sleeper support L (mm) under the rail seat for both timber and concrete sleepers
L
as
=
Rg,
(4.3)
where R and g are as previously defined. The quantity (Rg) used by both Clarke and Schramm in Equations 4.1 and 4.3 can also be expressed as twice the overhand length of the sleeper Q (Figure 4.2)
.
2) seat AS ( m m
The effective~sleepersupport area beneath the rail can now be calculated by the following equations: (a) According to Clarke (1957), for timber sleepers
(4.4) (b) According to Schramm
i
l
(1961), for both timber and concrete
sleepers
where B
=
sleeper breadth (mm).
If the sleepers are not of a uniform breadth; for example concrete sleepers manufactured a in manner where the centre breadth is less than the end breadth; it is reasonable to assume an average sleeper breadth over the effective of length sleeper support when determining the effective sleeper support area. For steel sleepers with an inverted trough shape crosssection it is reasonable to assume that the effective sleeper
breadth is the maximum distance between the sleeper flanges when
130
Figure 4.2 Principal sleeper dimensions
131
determining the effective sleeper support area because of the ballast confinding characteristics of the sleeper section. DETERMINATION OF THE MAXIMUM RAIL SEAT LOAD
The exact magnitude of the depends upon the following
load applied torail each seat known and unknown parameters:
.
. . . .
the rail weight the sleeper spacing the track modulus per rail the amount of play between the rail and the sleeper the amount of play between the sleeper and the ballast proud sleeper plates (in the case of timber sleepers).
.
The influence of the last three factors vary in accordance the standard of track maintenance, and their effect is to redistribute the applied sleeper loading to sleepers with adequate support.
wit
Various methods have been developed to calculate the magnitude of the rail seat load and of these, the methods outlined in detail in the following section are thought to be the most frequently used. The Beamon a Continuous Elastic Foundation Model
The maximum rail seat load can be determined theoretically by using the beam on a continuous elastic foundation model, and this is the approach suggested by Talbot (19181934) and by Clarke (1957). Using this model the maximum rail seat load qr (kN) can in general be determined by
132
where
S k ym
= =
= =
F1
sleeperspacing (m), track modulus (MPa) per rail, the maximum rail deflection caused by the interaction of a number of axle loads about a given reference position (mm) (Figure 3.13), and factor of safety to account for variations in the track support cassed by variations in the standard of track maintenance (Clarke (1957) adopts a value of F1 equal to l).
The Special Committee on Stresses in Railraod Track (Talbot 19181934) came to the following conclusions about how various combinations ofrails, sleepers and ballast react under load:
.
tests showed that for a given modulus of rail support the maximum rail depression is not greatly affected by the rail section, althoughit is slightly greater for lighter rail sections (Danzig et al. 1976) for a given wheel load, the product of rail depression and the modulusof rail support ( k y , ) is nearly constant, consequently the upward resisting pressure is nearly constant
.
and
is approximately the same for any rail section and any support stiffness included in the tests. From this it follows that the pressure of rails (i.e. the rail seat load) is also the same regardless of the weight of rail or value of the modulus of support (Danzig et al. 1976). O’Rourke (1978) has shown theoretically that under unit train conditions, where the adjacent wheel loads can be assumed to be the same, the product kym/unit load is, for a given rail section, nearly constant for any value of track modulus. It was found that this product was mainly dependent upon the axle spacing the vehicle, (smaller axle spacings having a more significant effect, as expected, referto Figure 4.3). For the heavy axle load conditionson the Mount Newman Mining Co. and the Hamersley Iron railways, a formula was developed to determine therail 4.6 to seat load qr(kN), which simplifies Equation
of
133
Note: Rail Moment of Inertia I = 37.5 X 106m4(90in4 )
Figure 4 . 3 Values of the modulus of rail support (k y) for various axle spacings and assumed track moduli (O'Rourke 1978)
134
qr where
S P F1
=
= = =
0.56 SFIP sleeperspacing(m), design wheel load (kN) , and factor to account for variations in the track support caused by variations in standard of track maintenance.
(4.7)
the
The coefficient0.56 represents the average value of the product kym/unit load (Figure 4.3), for the smallest axle spacings of the iron ore wagons.
These axle spacings being approximately 1.8 m (for axlesin the same bogie), and 2.3 m (for axles between adjacent wagons). The AREA method The AREA (1975) has developed a relationship to determine the maximum rail seat loadqr (kN) andis implicitly based upon the beam on elastic foundation model. Examination of Magee's Reports (19651971) ledto an understanding of how this method was developed. It was recognised that none of the experiments conducted by Talbot (19181934) gave comparative data on the effect of sleeper spacing on the rail seat load; all of Talbots' experimental work being based on a sleeper spacing of about 510 mm. As a basis in the development of a method to calculate the maximum rail seat load, the beam on the ealstic foundation method was adopted and a track modulus per rail of 13.8 MPa for a sleeper spacingof 510 mm was assumed. Therefore using Equation 3.27 the track spring constant D(kN/mm) for the rail support was defined to be
It was also recognised that the larger the bearinq of area the sleeper on the ballast, the more resistant the sleeper will to depression under loading. Tc account for this effect the
be
135
effective area of sleeper support beneath the sleeper was inco orated into the expression €or the track spring constant. This was achieved by assuming that the track spring constant o 7 kN/mm related specifically to a 2590 X 230 X 180 mm sleeper. As previously mentioned, Magee assumed that this sleeper has an 'area effective lengthof 990 mm and consequently an effective of 3 2 228.10 mm . The track spring constant D (kN/mm) was determined for any other sleeper size by the following assumption
228.10"
where
h,
=
effective2sleeper support area beneath seat (mm ), of any other sleeper.
the
rail
For a rail size of 57 kg/m with an adopted vehicle configuration of four 150 kN wheel loads spaced apart by ,1.8, 2.0 and 1.8 m respectively, the maximum rail deflection was calculated by the
beam on an elastic foundation method (using the above relations to determine the track modulus a for variety of sleeper sizes). The maximum rail seat load was then calculated using Equation
4.6. Magee (19651971) also found that the product of track modulus and the maximum deflection, k y , , is nearly constant and consequently that the influence of the effective area, Equation 4.8, did not havea significant effecton the magnitude of the
rail seat load. A simplified diagram was then constructed, based on the above vehicle loading and the on beam elastic foundation method, which shows that the rail seat load (express as a percentage of the wheel load) varies in proportion to the
4.4, sleeper spacing (Figure
line A).
It would appear that the AREA (1975) have based their method of determining the maximum rail seat load upon this diagram. Basically the AREA metnod can be expressed by
136
where qr
P
DF
= =
=
maximum rail seat load (kN), design wheel load (kN), and distribution factor, expressed as a percentage of the wheel load.
Recommended values of the distribution factor for timber, concrete and steel sleeper types and over a range of sleeper spacings are also presented in Figure 4.4; (lines B, C and D respectfully). It is clearly evident that AREA recommended distribution factors for timber and concrete sleepers (lines B and C) have identical slopes to Magee's estimate (line A). This leads to the conclusion that the AREA estimate of the distribution factor for timber sleepers, line B, is based upon Magee's estimate the line A being shifted upward on the vertical scale by 10 per cent. This would correspondto an equivalent factor of safety, F1 in Equation 4.6, of between 1.22 and 1.33. When prestressed concrete sleepers were first introduced into
the America1 railways networks in the early 1960s many inservice performance problems were encountered (Weber 1975). The AREA Special Committee on Concrete Ties (AREA 1975, Weber 1975) was established to investigate the causes of these performance problems and to recommend specifications for prestressed concrete sleepers which would reasonably guarantee reliable performance under mainline conditions. One of the Special Committee's recommendations was that the assumed distribution factor €or timber sleepers, line B, should for concrete sleepers be shifted upward a further 5 per cent on the vertical scale, resulting in line C. Upon comparison with the original Magee estimate, line A, the net effect is to increase the factor of safety F1 in Equation 4.6 to between 1.33 and 1.50. A distribution factor has been adopted for use with the of steel sleepers and this is shown as D line in Figure 4.4 (Brown and Skinner 1978). design
24722/8021
137
l
I
I
27.5
25
22.5
Figure 4 . 4 Approximate percentage of wheel load carried by an individual sleeper
13 8
The ORE !.lethod The ORE (1969) have developed a statistical method to calculate the maximum rail seat load on an individual sleeper. Based upon experimental data a formula was empirically derived for determining the maximum rail seat load qr(kN), i.e.,
qr where

E E
P
=
=
design wheel load, (kN) based upon the ORE formula for the impact factor, dynamic mean value of the ratio qr/ where and P S PS are the mean values of the rail seat load


ir
and
=
the the
static axle load respectively, maximum value of the ratio qr/Ps,


E
is approximately equal to 1.35.
Measured values of in Table 4.2.
L
are tabulated for various types of sleepers
The applied wheel load on the rail considered between three adjoining sleepers
as
being
distributed
Due to the effects of the unknown parameters it is commonly assumed that the maximum possible sleeper loading occurs when the wheelload on a railis distributed between three adjacent sleepers. The load at the rail seat of the central sleeper is denoted by X per cent and the load at the railseat of the two other adjacent sleepers is x/2 per cent. The valueof X is commonly assumed as 50 per cent of the design wheel load (equivalent to 25 per cent of the design axle load). This simple assumption has been used by Baymond (1971), Heath and Cottram (1966), Eisenmann (1969a) and by the ORE (ORE 1968). rail seat load qr(kN) is The maximum
139
TABLE 4.2
Sleeper
 VALUES OF THE RATIO E
= %/p
S
FOR VARIOUS SLEEPER TYRSAND SPACINGS (ORE 1969)
Type
Sleeper Dimensions Sleeper Area of Sleeper Q(a) length X breadth X spacing support rail seat thickness (mm) 2 3 (l0 X mm (mm)
~~
(mm)
E
=
9
r , $
S
(range)
400 0.56 0.32 450 550 0.32 0.46
French, type VW Prestressed Concrete British, type F Prestr.essed Concrete
2300 X 250 X 140 2515 X 264 X 760 200 2400 X 300 X 190 600 2600 X 255 X 135
600
200.0 268.8 510
252.0
 0.59
0.48
0. 44
, "
P
German, type 5 8 Prestressed Concrete French, Hardwood

600
280.5
 0.76
(a) The area of sleeper support = 2Q X sleeper width at rail seat, whereQ of the centre line of the rail from the end of the sleeper.
=
distance
qr
= =
0.5P
,
(4.11)
where
P
design wheel load (kN)
.
A comparison of the above methods used in the calculation of the maximum rail seat load is presented in Table 4.3. It is interesting to note that the value of the factor of safety F1 used by the AREA for concrete sleepers is approximately 1.33 to 1.50 whereas the ORE recommended value ofCl is 1.35. The determined value ofC l is based upon experimental data and would seem to more justifiable for use than the AREA assumed values of F1.
be
Method
3 Ad j acent Sleepers Method qr
Maximum Rail (qv) (kN)
L
Seat
Load
=
qr qr
0.5OP 0.43P (a) 0.6OP 0.65P
BEF Formula (O'Rourke 1978) (Mt Newman ore car axle spacing and sleepers at 760 mm centres)
AREA Method (Birmann 1968) (Prestressed concrete sleepers at 760 mm centres)
= = =
ORE Method (ORE 1969) (BR Type F prestressed concrete sleepers at 760 mm centres) qr
(a) In BEF Formula, a factor of safety (F1) to account for variations in the track must be applied to the value determined above. If F1 assumed to be 1.5, then qr =
0.65P.
NOTE: Where P = design wheel load (kN), and the sleeper spacing adopted is 760 (mm) for the comparison.
In the selection of the required sleeper size and sleeper spacing, the conventional design procedure as illustrated by Figure 1.1, is essentially based upon a trial and error method. A sleeper size is adopted and a sleeper spacing is selected on the basis of
141
that which satisfied the tolerable limits of sleeper to ballast contact pressure and flexural capacity. The methods outlined previously to calculate the effective sleeper support area and the magnitudeof the rail seat load are used in the following sections, which will discuss in detail:
. .
the contact pressure between the sleeper and the ballast the maximum allowable contact pressure the maximum sleeper bending moments and bending stresses the flexural requirements of various sleeper types.
. .
Using an adopted sleeper size in conjunction with any selected design method the maximum allowable sleeper spacing is that which just satisfies the above sleeper design constraints. THE The CONTACT general PRESSURE approach BETWEEN for the THE SLEEPER AND’THE BALLAST of the contact
calculation
pressu
beneath the rail seat is to assumea uniform contact,pressure distribution over the assumed effective area of the sleeper. This assumption is made in order to facilitate the ease of calculations. A factor of safety is then usually applied to account for variations in the sleeper support. The average contact pressure between the sleeper and pa(kPa) in all the developed formulae can be reduced to the
balla
where
qr
=
maximum rail seat load (kN), breadth of sleeper (m), effective length of sleeper under the rail seat (m), and factor depending upon the sleeper type and the standard of track maintenance.
B L
= = =
14 2
Clarke (1957) has used this method to determine the average contact pressure between the sleeper and ballast; (L being determined by Equation 4.1). Nowt assuming that the effective length of the sleeperL, is approximately a third of the total sleeper length L(m) using Equation 4.2, it is possible to rewrite Equation4.12 as P ,
=
[ z ]
F2
.
(4.13)
The AREA recommends that when calculating the average contact pressure between the sleeper and the ballast the maximum rail seat load should be doubled and the average contact pressure calculated for the full length of the sleeper. AREA The has adopted this value of two for F2 to account for possible excessive contact stresses due to non uniform sleeper support caused by the lack of track maintenance. The result of this procedure is to effectively determinea maximum contact pressure between the sleeper and the ballast for the condition of non uniform sleeper support. This contradicts the initial assumption of uniform support over the entire sleeper length. The AREA formula for calculating the average pa(kPa) between the sleeper and the ballast is contact pressure
a ' where
=
pgrl
F2
I
(4.14)
= totallengthofthesleeper(m) qr, B and F 2 are as above.
Raymond (1977) has suggested that the value of F2 equal to 2 has been adoptedby the AREA in order to account for the worst possible conditions of contact pressure between the sleeper and the ballast. This occurs when lateral forces are applieci at the rail seat and are caused by flanging of the wheels against the rail head whilst travelling through curves. In Figure 4.5 it is suggested that extremely high lateral forces act sirmltaneously
143
P
1.520m
P
" "
1
p
l
Pressure Distribution with Lateral Load
2P
R e c o m m e n d e d A.R.E.A. Design Contact Pressures for Sleeper Support
"
Figure 4.5 Effect o f lateral loads o n sleeper contact pressure and the A.R.E.A.recommended design assumption ( R a y m o n d 1977)
144
on both rails in the same direction. In travelling through curves contact of the wheel on the rail takes place at two points on one rail, (at the rail head and at the side of the rail head). This loading condition results a inhigh lateral force on one rail causedby the contact of the wheel flange with the rail, but a smaller lateral force on the other rail caused by the component of the wheel load acting at the top of the rail head. Consequently, it would based appear that upon the the value of the judgement safety of the factor ;ias been railway authorities entirely
and researchers. In fact Clark (1957) has not recommended a value of F2 in his paper. The assumed contact pressure distribution at the rail suggested for use by the AFCGA and referred to by Kerr (1976) and Schramm (1961) is shown in Figure 4.6. Both authorities assume that the contact pressure is uniform over the effective length L(m) given by L where g
=
=
R
g,
(4.15)
=
total length of the sleeper (m), and distance between rail centres (m).
This assumed presure distribution forms the basis of the sleeper moment calculationsat the rail seat used by both authorities. According to Kerr (1976) the AREA has also assumed a pressure distribution under the middle portion of the sleeper for use in the calculationof the maximum midspan sleeper moment, and this . is presented in Figure 4.6 (b) The ORE (1969) have carried out experiments to deternine the contact pressure between the sleeper and the ballast due to the action of known wheel loads passing over the track. Tests were carried out €or a range of:
.

sleeper types timber concrete 145
24722/8022
.
sleeper spacings

630 760
.

type of ballast packing condition under the sleeper manual portable electric hammer machine packed. to
The experimental values of the contact pressure were reduced normalised form pN(kPa/kN), by dividing the observed contact pressure po(kPa) with the rail seat load qo(kN) causing that contact pressure, i.e.,
(4.16)
Results indicate that the manual packing method produces 40 per cent higher than the pressures pressure values some observed with both the portable electric hammer and the mechanised packing methods.
conta
fully
The generaltrend of these results are as follows (ORE 1969):
.
the peak normalised contact pressures pN are in the range of 5 to 7 kPa/kN the maximum contact pressures occur in the vicinity of the rail seatand the minimumcontact pressures occur in the middle zone of the sleeper; and this is clearly illustrated in Figure 4.7.
.
The ORE concluded the following from these experiments:
(a) The sleeper to ballast contact pressure distribution is very much a function of the ballast packing method usedr and that it tendsto be, "unpredictable and random with a high degree
of scatter".
14 6
I
Q g
I
pa =
4r
BL
[Effective Length]
Figure 4 . 6 ( a) Assumed distribution for the determination of the a i l seat as used by maximum sleeper bending moment at r Battelle (Prause et 1974) al the area (Kerr 1976) and Schramm (1961)
I
fi
I
Contact Pressure Pa
I
1
1
I
Figure 4 . 6 ( b) Assumed distribution f o r the determination o f the maximum f the sleeper, as used sleeper bending moment at centreo by Battelle (Prause a l et 1974) and the area (Kerr 1976)
Assumed contact pressure distributions between sleeper and ballast f o r sleeper bending moment calculations
14 7
South Rail Legend:
North Rail
use in calculations of  Assumed idealised contact pressure distribution for stresses (ORE1968,Figure 6 d )
1 1 3 Effective support length under each rail approaches 1
Note: Thickness of construction 230 m m (9in) Concrete sleepers at 790 m m (31 in) Portable electric h a m m e r packing M e a n of two passes of two axles Force balance error = 9.0%
Figure 4 . 7 Measured sleeper soffit to ballast contact pressure for concrete sleepers and a n assumed idealised contact pressure distribution (ORE1968)
14 8
A s the
contact pressures developed under timber sleepers are of the same order of magnitude as those beneath concrete sleepers, it was considered that flexural rigidity of sleepers wasa secondary influence on the magnitude of the contact pressure, (i.e. sleeper size and spacing having a far more significant influence).
Sleeper spacing does not appear to influence the contact pressures when defined in a normalised form although absolute values of contact pressure vary with the sleeper spacing. to the fact that the contact pressures under the sleeper are
of a random nature the ORE has suggested that an "equivalent" uniformly distributed pressure distribution in the region under the rail seat be used in the calculations of the sleeper bending moment, (Figure 4.7) . Using this assumed pressure distribution it was observed with concrete sleepers,2.510 m long and 0.260 m wide, that the maximum normalised contact pressure between the sleeper and ballast was
pN
the
=
4.75 kPa/kN ,
i.e. 4.75 kPa of contact pressure between the sleeper and the ballast for every 1 kN of rail seat load.It is evident that this observed value of pN is dependent upon the sleeper size. The maximum allowable contact pressure
The recommended limit of the maximum allowable bearing pressure between the sleeper and the ballast varies considerably with railway authorities. All the suggested limits do not make allowance for changes in sleeper size or variations in sleeper support. It also is not clearly apparent whether these suggested limits are based on experimental data of ballast crushing or on an arbitrary selection of a likely limit.
149
Combining Equations 4.9 and 4.14 and using the recommended value of F2 equal to 2, the AREA formula used to calculate the allowable bearing pressure between the sleeper and the ballast pa(kPa) for timber sleepers is
a ' where P DF
B
Q
=
= =
 4 P(DF)
BR
(4.17)
design wheel load (kN) (i.e. static wheel load multiplied by the AREA impact factor), the AREA distribution factor (Figure 4.4 line B) for timber sleepers, breadth of sleeper (m) , and total length of sleeper (m).
=
Using this equation, the AREA design manual recommends €or timbe sleepers a maximum allowable contact pressure between the sleep and the ballast of 450kPa, (65 psi) (AREA, 1973). For the case of concrete sleepers the AREA Special Committee on Concrete Tie (AREA 1975, Weber 1975) has specified that the contact pressure between the sleeper and the ballast pa(kPa) be calculated using the formula (4.18) where
PS
DF
= =
=,
static wheel load (kN) , the AREA distribution factor for concrete sleepers (Figure 4.4 , line C) , impact factor, the AREA assumed value for all conditions is 1.5, breadth of sleeper (m) , and total length of sleeper (m).
~
6
B
Q
= =
The significance of the above AREA procedures for determining the contact pressure between the sleeper and the ballast will now discussed. Assuming that the impact factor used in Equation 4.17 has a value of 1.5, and that both timber and concrete sleepers used have identical dimensions. For a sleeper spacingof 600 mm the AREA distribution factors obtained from Figure 4.4 for timber and concrete sleepers are 0.45 and 0.5 respectively. Upon 150
substitution of the above values into Equations 4.17 and 4.18 it is apparent that the magnitude of the contact pressure between the sleeper and the ballast obtained are approximately the same. Notwithstanding this observation, it should be emphasised that the AREA recommended design procedure for concrete sleepers significantly increases the design bending moment capacity of the sleeper at both the rail seat and centre region.
When using the above equation the AREA recommends that the average contact pressure between the sleeperthe' and ballast should for concrete sleepers not exceed 390 kPa, (85 psi). This limit was suggested for highqualityabrasionresistant ballast and if lower quality ballast materials are it used should be reduced accordingly (1). The above recommendations of the maximurn allowable contact pressure between the sleeper and the ballast are all based factored values of the calculated maximum rail seat load. When using Equation4.12 to calculate the contact pressure between the sleeper and the ballast; Clarke(1957) has recommended a maximum allowable limit for timber sleepers of 240 kPa (35 psi). This limit is based on an unfactored average contact pressure. Eisenmann (1969a) has also stated that special steps should be taken to ensure that the bearing pressure between the sleeper and the ballast does not exceed 300 kPa. This limit also appearsto be based on the effective area of sleeper contact, Equation 4.12, and the apparent unfactorinq of the rail seat load. The differences between the above recomendations are entirely due to the design assumptions used in the alternative methods. The AREA use a factored rail seat load value and calculates the uniform average contact pressure for the full sleeper length, whereas Clarkeand Xisenmann an use unfactored contact pressure rail seat load,but calculate the uniform average for the effective length under the rail seat.
upon
(1) The AREA do not suggest any iaethod to ,deterrr,ine =;?is reduced contact pressure limit.
151
Using an adopted sleeper size in conjunction with of the any above design methods a first estimate of the maximum allowable sleeper spacing is obtained, and is that which satisfies the recommended limits of the allowable contact pressure. The preliminary sleeper size together with the trial sleeper spacing is further analysed in order to limits are also not exceeded. THE CALCULATIONOF THE MAXIMUM check that the sleeper
flexura
SLEEPER
BENDING
MOMENTS
As previously
mentioned the adopted sleeper size together with
the first estimate of a trial sleeper spacing is further analyse to check the sleepers flexural performance under the service load conditions. The maximum bending moments and bending stresses occur at the following locations along the sleeper length:
.
.
at the region of the rail seat at the centre region of the sleeper.
Consequently the following sections will analyse the sleeper flexural performance and requirements at these two locations.
Currently, sleepers are manufactured from the following engineering materials: timber, (both hardwood and softwood), concrete (both prestressed and reinforced) and steel. The analysis of the maximum sleeper bending at the rail seat and at the centre of the sleeper is identical regardless of sleeper material used, but the type of flexural limits imposed upon the sleeper varies according to the sleeper material type. Thus the flexural limit for both timber and steel sleepers is essentially the allowable bending stress, whereas for concrete sleepers, notably of the
prestressed type, the flexural limit is usually expressed in the form of a positive or negative bending moment capacity.
152
The
maximum
sleeper
bending
moment at the
rail
seat
The fundamental basisof the methods proposed by Clarke (1957), Schramm (1961) and Battelle (Prause et al. 1974) for determining the maximum sleeper bending moment at the rail seat is the assumption of a uniform effective contact pressure distribution between the sleeper and the ballast. Kerr (1976) states that calculations based upon this simple assumption yield upper bound solutions for the expected sleeper bending stresses and are therefore considered sufficient for sleeper design purposes. The following are the commonly used design rtethods for determining the maximum sleeper bending moment at the rail seat. Solution according to Battelle: Batelle (Prause et al. 1974) has indicated that the equations derived by Clarke (1957) for determining the bending moment at the rail seat are dimensionally inconsistent, (the assumed uniform distributed load was inappropriate. The following formulations have been suggested by Battelle for use in the analysis of sleeper bending moments: (a) The upper bound solution for the maximum sleeper bending moment at the rail seat can be calculated by considering the sleeper is in the "endbound'' condition. Thus the entire support offeredby the ballast to the sleeper is considered as a point load, equal to the rail seat load, at located the end of the sleeper, Figure4.8(a). The solution for the maximum bending moment at the rail seat Mr(kNm) is the simple couple (4.19) where qr g
=
=
=
maximum rail seat load (kN) , overall sleeper length (m), and distance between rail centres (m).
24722/8023
153
(b) A less conservative and probably more realistic solution is recommended by Battelle (Prause et al. 1974). This solution assumes that half the rail seat load is distributed over the sleeper overhang length (Qg)/2 Figure 4.8(b). The maximum sleeper bending moment at the rail seatMr(kNm) is therefore (4.20) where
g , r
R and g are as previously defined.
The above solutions are independent of the type of sleeper material used in the analysis. Solution according to Schramm: For sleepers fitted with sleeper bearing plates the effect that the plate on has the calculations of the maximum sleeper bending moment at the rail seat has been considered by Schramm (1961), Figure4.8(c). The maximum sleeper bending momentat the rail seat Mr(kNm) for this condition can be calculated by (4.21) where
j
=
gr ’ The maximum
R and
length of sleeperbearing g are as previously bending
and defined. sleeper
sleeper
moment at the centre of the
The maximum sleeper bending moment at the centreof the sleeper occurs when the sleepers are said to be centrebound. This leads to a problem in defining what the contact pressure distribution is for the centrebound condition. Talbot (19101934) observed that forthe centrebound condition the pressure distribution between the sleeper and the ballast approached a uniform distribution (Figure 4.1). The following are the commonly used design methods for determining the maximum sleeper bending moment at the centre of the sleeper.
15 4
I
I
q,
I
I
Sleeper
9,
Contact
Pressure, Pa
I
L
Upper
pg
+
Figure 4 . 8 ( a) boundsolution (Prause et al 1974)
I
l
!
Sleeper
Contact Pressure, Pa
I
L
I
'
Qg
I
I
I
I
Figure 4 . 8 ( b) Less conservative solution (Prause et al 1974)
l*
Solution for
Qg
4
Figure 4 . 8 ( c ) sleepers fitted w i t h bearing plates (Schramm 1 9 6 1)
Estimation of maximum sleeper bending m o m e n t at the rail seat
155
Solution according to Battelle: According to Battelle (Prause et al. 1974) the bending moment at the centreof the sleeper can be calculated by assuming that the sleeper is centrebound. Battelle represents this condition by assuming the sleeper is resisted by a single point load at the centre of the sleeper, and the maxim negative bending moment at the centre of the sleeper M (kNm) can
C
be calculated from (4.22) where qr g
= =
maximum rail seat load (kN), and distance between rail Centres (m).
This solution proposed by Battelle can only be regarded as extremely unlikely to occur in practice. If the support condition, (and the implied contact pressure) under the sleeper reached t state due to the lack of track maintenance, the load on the sleeper would most probably be redistributed to other sleepers. Nevertheless this expression can be regarded as being the uppe bound solution for the maximum sleeper bending moment at the centre of the sleeper. Solution according to Raymond: Raymond's
(1977) definition of
the assumed centre bound condition is based upon earlier exper mental work carried out by Talbot (19181934). The solution for" the maximum bending moment at the centreof the sleeper is therefore calculated on the basis of an assumed uniform bearing pressure over the total length of the s l e e p e r ' ' ) . The maximum negative bending moment at the centreof the sleeperM,(kNm) can be calculated from
(1) For this assumption the uniform pressure between the sleeper and the ballast p(kPa) is
p = 2qrm where B = sleeper lxeadth (m).
156
(4.23)
where
qr g
= maximum rail seat load (kN),
= =
distance between rail centres (m), and total sleeper length (m). CAPACITY AND REQUIREMENTSOF SLEEPERS
THE
FLEXURAL
As previously mentioned the type of flexural limit imposed upon the sleeper varies according to the sleeper material type used in the track design. The flexural limits of timber and steel sleepers are based upon an allowable bending stress limit. Whereas with prestressed concrete sleepers the flexural is limit entirely based upon the designed bending moment capacity of the manufactured sleeper. In the following sections the current methods of analysis used to determine the flexural performance of timber, prestressed concrete and steel sleepers are outlined. The Flexural Requirements of Timber Sleepers With timber modulus
The maximum sleeper bending stress at the rail seat:
sleepers the cross sectional shape is rectangular and uniform along the entire sleeper length consequently the section 3 of the sleeper Z (m is B t ' z = 6 where
,
(4.24)
B t
= =
sleeper breadth (m), and sleeper thickness (m). stress
Therefore the upper bound estimate for the sleeper bending at the rail seat oru(MPa) can be calculated from equations 4.19 and 4.24 and is
15 7
0
ru
 Mr
Z
"

3.4,
(Qg) 103 Bt2
(4.25)
'
where qr
L
=
4 B
t
rail seat load (kN) , total sleeper length (m) , distance between rail centres (m), sleeper breadth (m) , and sleeper thickness (m).
= = =
=
Clarke ( 1 9 5 7 ) uses a similar approach but states that the rail seat load determined using the beam on an elastic foundation
analysis (Equation4.6 and F1 = 1) should be doubled when calculating the sleeper bending stress at the rail seat in order to allow for variations in the effective sleeper support condit Consequently if Clarke's approach was followed Equation 4.25 would be multiplied by 2.
of Using Equations4.20 and 4.24 a less conservative estimate the sleeper bending stress the at rail seat or(MPa) is
I
(4.26)
where qr
I
Q,
g, B and t are as previously defined.
at Using Equations4.21 and 4.24 the sleeper bending stress the rail seat or(MPa) for sleepers fitted with bearing plates i (4.27) where j
=
length of sleeper bearing plate
(m), and
9 , L , g, B and t are as previously defined.
It is apparent that the sleeper bending stress at the rail seat can be reduced considerably when sleepers are fitted with bear plates. For example, consider a standard gauge track comprising of 54 kg/m rails laid upon2.440 m long timber sleepers. Using identical valuesof the rail seat load the sleeper bending
15 8
stress at the
rail
seat
can
be
reduced by roughly 30 per are fitted between
cent the rail
when 0.300 m wide and the sleeper.
sleeper
plates
The maximum sleeper bending stress at the
centreof the sleeper:
Depicting the sleeper centrebound conditions as being represented by an assumed uniform pressure distribution over the total sleeper length, the maximum bending stress at the centre of the sleeper oc'(MPa) can be calculated using Equations 4.23 and
4.24 as follows: (4.28)
where
qr L !
9 B
= = = =
rail seat load (kN), total sleeper length (m) , distance betweenrail centres sleeper breadth (m), and sleeper thickness (m). tensile bending stress of timber sleepers: (m) ,
t The maximum
=
allowable
Notwithstanding the effects of non uniform properties, the maximum tensile bending stress of timber sleepers varies predominantly with the following considerations:
.
. .
the timber type, which in general terms can be classified as either a hardwood or a softwood
whether the sleepers are treated or untreated with preservatives the timber moisture content.
Clarke (1957) recommends that the maximum tensile bending stress for timber sleepers should not exceed 5.5 MPa, whereas the limiting value recommended by the AREA (1973) is 7.6 MPa. It should be noted that both Clarke and A W the A 60 not makeany distinction between types of tis3er sleelsezs. The Anerican of sleepers are typically of timber types used in the manufacture
139
the softwood variety, and it would therefore appear reasonable to assume that the AREA recommendation applies to softwood sleepers Battelle (Prause et al. 1974) states that the maximum allowable tensile bending stress recommended by AREA the corresponds to a
value determined for the lowest grades of oak and pine sleeper which had been subjected to long duration loading and also exposed to wet conditions coupled with a moderate decay hazard
Battelle also states that the endurance limit of sleepers subjected to cyclic loading has been experimentally determined to be 28 p cent of the modulus of rupture, Erupt, (for American timber types). Experimentally determined values of the modulus of rupture for Australian timber types have been quantified by Duckworth (1973) and Reid (1973) and these are presented in Tables 4.4 and 4.5. Also in these tables is presented an estimate of the sleeper endurance limit based upon the experimental of modulus of rupture values of new sleepers. The variation sleeper strength with the condition of the new hardwood sleeper 4.4. Upon comparison of the experimentally is presented in Table determined ultimate strength of treated and untreated sleepers it
is noticeable that the treated sleepers show a narked reduction in flexural strength. TABLE 4.4
 EXPERIMENTALLY
1973
DETERMINED VALUES OF THE AVERAGE MODULUS OF RUPTURE AND ESTIMATES OF THE ENDURANCE LIMIT FOR VARIOUS NEW HARDWOOD SLEEPER CONDITIONS (DUCKWORTH
Hardwood Sleeper Conditions
Average Modulus of Estimated Endurance Limit (28 per cent of Rupture (from 12 tests) Erupt W a ) Erupt) (MPa)
110 80 78
l
Dry untreated Green untreated Green treated and incised Dry treated Green treated
32 23 22 17 13
61 47
160
m
N
e
7
TABLE 4.5
 EXPERIMENTALLY
DETERMINED MODULUS OF RUPTURE VALUES FOR VARIOUS TIMBER SLEEPER TYPES AND AGE CONDITION TOGETHER WITH AN ESTIMATE OF THE ENDURANCE LIMIT OF THE NEW SLEEPERS (REID 1973)
Sleeper Type and Sleeper Sleeper Cross Section Modulus of Rupture Endurance Limit Width Condition Age Erupt X Depth Sleepers of New (28 per cent (Erupt) (years ) (mm) (MPa) (mal New Sof twood New Softwood
P
p
of
34 233 22 230 37 37
X X
173 170
10 6
cn
Softwood Softwood Hardwood Hardwood Hardwood Hardwood (oneoutsidespike) Hardwood Hardwood (two outside spikes) Hardwood (two outside spikes
238 X 165 236 X 166 257 X 133
20 20 55 23 19 46 34
23 29
New
37 37 17 17 17 17
255 267
X X
128 127
235 X 117 248 X 117
251
251
X
X
114
121
It is significant that the estimated endurance limit of an Australian softwood sleeper, (Table 4.5) is of the same order of AREA recommendation of maximum allowable tensile magnitude as the bending stress. It is also apparent that the estimated endurance 4.5) is still less than the limit of the hardwood sleeper, (Table
modulus of rupture values of hardwood sleepers that have been subjected to 37 years of service life under secondary line conditions with wetting and drying and decay hazard. The reduct of the sleeper strength caused by respiking operations is also shown in Table 4.5. It should be noted that the condition of aged timber sleepers also dependent upon the cumulative volume of traffic carried at a given time. Under mainline conditions it is therefore anticipated that the life of the sleeper would be governed by other service factors before the sleeper strength is inadequatetodue age condition alone. The maximum sleeper bending stress determined at the rail seat
and the centre of the sleeper is compared with the allowable tensile bending s t r e s s ' ' ) . If for a particular adopted sleeper size this limiting stress is found to be exceeded, the firs,t estimate of the sleeper spacing is further reduced until the limiting condition is satisfied. If the required sleeper spacing determined is considered impractical another sleeper size is selected and the entire sleeper design is recommenced. The calculation of the minimum required timber sleeper
thickne
The minimum required sleeper thickness for timber sleepers can readily be calculated by rearranging the equations which determ the maximum sleeper bending stress at the rail seat and the centre of the sleeper. Into these rearranged equations the
( 1 ) It is apparent that the current limits of the allowable tensile bending stress of timber sleepers have been established primarily for the area of the rail seat.'
16 2
maximum
allowable
ballast
contact
pressure
and
the
maximum
allowable sleeper tensile bending stress are substituted for the effective contact pressure and the actual sleeper bending stress respectively. Thus the sleeper thickness obtained is the minirnum which satisfied the strength criterian. The final sleeper thickness is determined by adding to this minimum thickness an allowance for weathering and decay deterioration.
The minimum required sleeper thickness at the rail seat for sleepers without bearing plates fitted between the rail and sleeper: The minimum sleeper thicknesstmi,(m) at the rail seat can readily be calculated by rearranging Equation 4.26 and substituting valuesof the maximum allowable contact pressure
the
between the sleeper and the ballast pall(kPa) and the maximum allowable sleeper tensile bending stress cB(MPa) for the calculated values of the bearing pressure and sleeper bending stress respectively. Since the maximum allowable contact pressure between the sleeper and the rail seat canbe calculated by either the AREA method or by Clark's method therefore two solutions minimum required sleeper thickness can be stated. of the
The AREA solution: Rearranging Equation 4.14 the equation for maximum allowable rail seat load qmax(kN) becomes
(4.29)
where
Pall(ll= AREA limit of allowable contact pressure (KPa)
B L ! F2 F2
= = = = =
450 (kPa),
sleeperbreadth(m) , sleeper length (m), and track maintenance factor, according to the AREA 2.
Equating qmax to qr in Equation4.26 and rearranging, the minimum required sleeper thickness tmin(m) can be calculated from
163
(4.30)
where
aB
= =
allowable sleeper tensile bending stress and distancebetweenrailcentres,and R and F2 are as previously defined.
(MPa),
9
' a 1 1 (1)1
The Clarke solution: A s previously mentioned, Clarke's equation for bending moment is dimensionally inconsistent. Notwithstanding this fact, the equation proposed by Clarke to estimate the maximum sleeper bending stress at the rail seat is of the same form as Equation 4.25. It should also be noted that when determining the maximum sleeper bending stress Clarke suggests that the rail seat load should be multiplied by a suitable factor to account for variations in sleeper support. Rearranging Equation 4.12 the maximum allowable rail seat load qma,(kN) becomes (1)
(4.31)
where Clarke's limit b
R
g
of allowable contact pressure (kPa)
240 (kPa),
F2
= = = = =
sleeper breadth (m), sleeper length (m), distance between rail centres, and track maintenance factor, according to Clarke, F 2 = 2. in Equation 4.25 and rearranging, the minimum
Equating qma, to qr
required sleeper thickness tmin(m) can be calculated from
(1) Assuming that the effective length L
=
(Rg).
164
(4.32)
where
=
allowable sleeper tensile bending stress, as previously defined.
and
' a 1 1 (2)I k r g and F 2 are The minimum required sleeper sleepers with bearing plates
thickness at the rail seat for fitted between the and rail the
sleeper: Using a similar approach to the AREA solution for sleepers without sleeper plates, and rearranging Equation 4.27 the minimum required sleeper thickness at the rail seat tmin(m) can be calculated from
(4.33)
I
where
length of sleeperbearingplate (m), and F2 are as previously defined. pall(l)l 1 1 a ~ and J required sleeper thickness at the centre of the
j
=
The
minimum
sleeper: The minimum required sleeper thickness at the centre of the sleeper can be calculated by rearranging the solution for the 4.28 and submaximum bending stress determined by Equation stituting values of the maximum allowable contact pressure between the sleeper and the ballast pall(kPa) and the maximum allowable sleeper tensile bending stress ag(MPa) for the calculated values of the bearing pressure and sleeper bending stress respectively. Using a similar approach to the AREA solution above the minimum required sleeper thickness tmin(m) at the centre of the sleeper can be calculated from
0.5
I
(4.34)
where
a I g,
aB and F 2 are as previously defined.
165
If the
area
approach
to
the
minimum
required
sleeper
thickness
were adopted, the value used would be the larger of the solutions to Equations 4.30 or 4.33 and 4.34. The Calculation of the required prestressed concrete sleepers minimum flexural capacity of
Theoretically the equations developed to calculate the maximum bending momentat the rail seat and at the centre of the sleeper are equally valid for timber concrete and steel sleepers. In the following section two current methods specifically developed for the calculation of the required minimum flexural capacity of prestressed concrete sleepers will be presented. These methods being the AREA design method (AREA 1975) and the ORE design method (ORE 1969). The AREA Design Method: At the basis of the AREA design method recommended by the Special Committeeon Concrete (AREA 1975) is the assumed contact pressure distribution between the sleeper the ballast. The AFLEA state that over a period of time, because of repeated loads, vibration and crushing of the ballast, the ballast will gradually compact, moving away from the areas of greater concentration. The sleeper therefore settles slightly of the sleeper to into the ballast, allowing the centre portion resist someof the applied load, thus reducing the amount of load carried by the sleeper ends. Due to this gradual redistribution of contact pressure, the AREA design method assumes a uniform contact pressure distribution occurs over the entire sleeper length. This support condition produces positive flexure at the rail seats and negative flexure at the centre of the sleeper.
an
The approach of determining the required flexural capacity of prestressed concrete sleepers is consistent with the AREA method of determining the rail seat load, Equation 4.9 and the AREA method of determining the contact pressure between the sleeper and the ballast, Equation 4.18. The rail seat load gr(kN) thus determined is expressed as 166
where
p ,
DF
= = =
static wheel load (kN) , distribution factor, expressed as a percentage of the wheel load (refer to Figure 4.4, line C), and impact factor (the AREA assumed value for all conditions is 1.5).
#
Therefore the assumed uniformly distributed load W (kN/m) over the entire sleeper length R (m) is
W
= 2qr . R
(4.36)
It should be noted that in Equation 4.35 the effective assumed value of the impact factor is 2.5. One of the main reasons why
U.S.
impact factor was significantly increased was that many early prestressed sleepers had tendon bond length failures in the region of the rail seat (Weber 1973). The Special Committeeon was to
the
Concrete Ties (AREA 1975) recommended that of one the measures that should be used to alleviate this performance problem design the prestressed design load. concrete sleeper to resist a much
higher
According to the AREA Special Committee on Concrete Ties the flexural design requirements for prestressed concrete sleepers can be calculated from the following equations (AREA 1975): (a) The maximum positive sleeper bending moment at the rail seat M , (kNm) is given by
(4.37) where
2
=
g W
=
=
totalsleeperlength (m), distance between rail centres (m), and assumed unifornly distributed load (kN/m).
(b) The maximum negative sleeper bending moment at the centre of the sleeper Mc(kNm) is given by
MC
= wg8
2
Mr Equation4.38 to
(4.38)
Substitution 4.23, i.e.,
of
Equation 4.37 reduces
Equation
(4.39)
where
qr,R and g are as previously defined.
The bending moments calculated by this method can be regarded as being adequate to insure that the prestressed sleepers will crack under normal service conditions. The required flexural capacity of various prestressed concrete sleepers to meet the AREA design method are presented in Table 4.6. It should be noted that the AREA method of determining load is implicitin this tabulation. the maximum rail seat
no
The ORE Design Method: The ORE design method (ORE 1969) is based upon recorded experimental data on the performance of various prestressed concrete sleepers under actual track and laboratory conditions. From these experiments an empirical design method was established to determine the minimum required flexural capacity of prestressed concrete sleepers. According to the ORE, the maximum rail seat load qr(kN) can be expressed as qr where
P
= 5
 ClP, 
(4.40)
= design wheel load, based upon the ORE formula for factor, impact the 5 = dynamic mean value of the ratio q r ’ P S (Table 4.2), and Cl = E/: = 1.35.

16 8
TABLE 4.6
 FLEXURAL
PERFORMANCE REQUIREMENT FOR PRESTRESSED MONOBLOCK CONCRETE SLEEPERS (AREA, 19 75) Required Flexural Capacity (kNm) Without Rail Seat + Rail Seat (c) Centre Cracking Centre +
a Length ( ) Spacing (b) (m) (mm)

2.44
2.5122.6 24.9
0.7
533 610 686 762 533 13.0 610 686 13.0 762 533 13.0 22.6 610 13.0 686 762 533 610 686 762
24.9 24.9 24.9 24.9 25.4 26.6 28.3 29.4 25.4 28.3 31.1 33.9 31.1 36.7 31.1 33.9
13 .O 10.2 13.010.2 13.010.2 13.010.2 13.0
13.0
22.6 24.9 27.1 29.4
10.2
10.2
23.7 25.0 11.3
10.2 2.59 22.6 10.2
13.0 13.012.4
13.0 12.4 13.0 13.6 13.013.6
22.6 22.6 22.6 22.6 22.6
11.3
13.0 2.67 28.3 33.9
11.3 2.74 22.6 533 13.0 11.9 22.6 610 13.0 36.7 686 4.1 22.6 762 13.0
13.0
13.0
22.6
39.6
(a) Constant top width across sleeper. Reduced bottom width at centre of sleeper Increase rail seat and centre positive flexural requirementsby 10 per cent. Reduce centre negative flexural requirementby 10 per cent.
(b)
Interpolate for intermediate sleeper spacings. to 4.0
X
(c) Based on elastic fastenings with upward spring rate 2.26 of
106 N/m.
The ORE experiments indicated that the actual stresses in the sleeper during dymanic testing are higher than the theoretically determined stresses. This dynamic stress increment also varies according to the location in the sleeper i.e., the dynamic stress increment at the rail seat differs from that at the centre of the sleeper. Due to the empirical nature of the ORE design formulae
24722/8025
169
a sleeper parameter was chosen to adjust the theoretical bending moment calculation to the observed bending moment in the sleeper. The parameter chosen to adjust the bending moment calculation wa the "lever arm'': half of the sleeper end distance Q(m), reduced
by half of the lengthb(m), (at the level of the neutral axis of the sleeper) which can be determined by assuming a 45O load
4.9).
spread of the rail seat load through the sleeper material (Fig
This lever armA (m) is therefore b A = Q 2 ' The theoretical bending moment at the rail seat readily be calculated using this lever arm
(4.41)
Mr(kNm) can 4. , 9) , i.e., (Figure
The amount required to factorize this theoretical bending moment thereby enabling the calculation to satisfy the experimentally measured maximum bending moment, was determined empirically. The maximum "factorized lever arm" Am corresponding moment at the rail to the due to
maximum dynamic
observed sleeper bending effectswas stated to be
seat
Am
where
@
= @A =
(4.43)
coefficient of the increment of the sleeper bending moment at the rail seat due to dymanic effects.
The maximum valueof the empirical coefficient Q, was found to approximately equal1.6 when the observed sleeper bending moments at the rail seat were analysed. Combining Equations 4.40, 4.42 and 4.43 the maximum sleeper bending moment at the rail M , (kNm) canbe empirically calculated from seat
170
Similarly an empirical coefficient 9 relating to the increment of the bending stress in the central section of the sleeper due to dynamic effects was also determined from experimental observations. The value of E was found to approximately equal 1.2. The maximum sleeper bending moment at M , (kNm) was determined empiricallyto be MC the centre of the sleeper
(at the centre of the sleeper) .  Mr E1 2
E1 (at the rail seat)
(4.45)
The main problem with the ORE design method is that there is no adequate theoretical backing and therefore no real of way determining The improvements in the sleeper steel design and performance.
flexural
requirements
of
sleepers
~
Several types of metal sleeper have been developed principally from steel and cast iron (e.g. the I beam and Pot types respectively). Modern steel sleepers have an inverted trough shape crosssection which permits efficient use of material and provides for excellent mating between the sleeper and ballast structure. There are two fundamentally different types of trough sleeper, namely, those of uniform wall thickness and those with crosssectional shape comprising a thick "top", or web, thin "legs" and bulbous lltoesll. The former is manufactured by pressing flat plate and the latter by hot rolling the final shape. The latter method permits more nearly optimum use of material and is particularly suited to heavy sleepers where cost the saving in material outweighs the saving due to simplicity of the other method of manufacture. Rail fasteners suitable for steel sleepers comprise two categories: those with a rail base plate and those without. The former type are usually attached to the sleeper by welding and the latter
l? 1
Note: Lever Arm A
=
Qb
~
2
Figure 4 . 9 Determination of "lever arm"h (ORE 1969)
172
either by welding or by notches and tunnels pressed in the sleeper. The use of high performance bolts is not favoured due to the relatively higher cost of the bolt compared to an equivalent weld (Brown & Skinner 1978).
A tabulation of proposed BHP steel sleeper section properties suitable for heavy, medium and light axleload conditions is presented in Table 4.7.
Using the equations developed to calculate the maximum sleeper bending moments'at the rail seat and at the centreof the sleeper, the maximum tensile and compressive sleeper bending by simple stresses at each of the locations can be determined applied mechanics. At the rail seat the r,aximum compressive bending stress occurs at the web of the steel sleeper section. Whereas for an assumed centrebound condition the maximum compressive bending stress occurs at the toe of the steel sleeper section. bending
Thus at the rail seat the maximum sleeper compressive stress f, occurs at the web and is calculated by
(4.46)
where
M ,
= =
z ,
maximum sleeper bending moment at the rail seat (kNm) (from Equation 4.20) , and 3 steel sleeper section modulus about the web ( m m ); tensile bending stress .ft occurs at
whereas the maximum sleeper the toe and is calculated by Mr f t  r where Zt
106
(4.47) Zt
3
=
sleeper section modulus about the toe (mm ).
173
TABLE 4 . 7
 SECTION PROPERTIES OF
Design Axleload (kN)
PROPOSEU RANGE OF BHP STEEL SLEEPERS Suggested Sleeper Spacing
(mm) 660
Sleeper Application
P
4
Mass
Thickness of Web of Trough
(mm)
12
10 8
Toe to Toe Width B
(m)
W/m) 35
30
Overall Depth from Section Web to Depth Neutrzl D Axis y (mm) (m)
120
Moment of Inertia I
Section Modulus About W e b About Toe zW zT 106 (mm4) (mm3) 103 103 ( m m 3 ) 7.00
6.00
P
Heavy Haul Sleeper Mainline A Sleeper Mainline B Sleeper Secondary Track Sleeper
300
250
300
300
40
40
710
200 150
730 760
25 15
250
200
3.50
120 105
90
175 150
100
85 75
50
R
35 30
1.30
45
21
For the centrebound conditions of the naxinum sleeper bending stresses are also calculated at the centre of the sleeper. The maximum sleeper compressive bending stress fc occurs at the toe and is calculated by
where MC
=
maximum sleeper bending moment Equation 4.2 3 ; tensile bending
at the centre (kNm)
the
maximum
sleeper
stress ft occurs at the
web and is calculated by
(4.49)
where
MC, Zt and Zw are as defined above.
In order to ascertain whether a selected steel sleeper will be adequate for the determined design stress levels, these stresses must be compared with allowable limits that are based upon fatigue considerations. The life of a steel sleeper is dependent principally on its fatigue performance and this be may characterised by the performance of that part of the sleeper iswhich most susceptible to fatigue damage, namely the rail fastener. Brown and Skinner (1978) have suggested a 2rocedure for the design of steel sleepers together with tie associated design of the rail fasteners and the rail insulation requirements. With rail fasteners such as for example the welded shear connector type, they suggest that the maximurn allowable bending stress (both tensile and compressive) in the regior. of tie rail fastener should not exceed 69 MPa if the life of the fascener is to be at least the life of the sleeper. At all cmther loza+; lL~ns of the parent steel an allowable fatigce stress 25 1512 ? : P O , (5otk tensile and compressiblej is sugcjested.
CHAPTER 5
 BALLAST
ANALYSIS
The primary functions of the ballast layer are (Prause et al. 1974, Robnett et al. 1975):
.
to provide a firm, uniform bearing surface for the sleepers which can be tolerated by the subgrade material, thereby limiting excessive differential settlement and the resulting
loss of vertical
and to transmit the imposed track loadings at a pressure lev
track
geometry
.
to provide the necessary lateral and longitudinal stability to the track structure thereby enabling it to resist the imposed vehicular loadings in curves and the thermal forces developed by continuous welded rail to facilitate track maintenance operations, such as the correction of track surface and alignment errors
.
to provide adequate drainage of the track structure, draining retard the possible growth of vegetation.
water away from the loaded zone of the subgrade and also to
As the current practice €or designing conventional railway track of the individual is based upon satisfying the strength criteria track components, the design of the required ballast layer depth is therefore based upon that depth which reduces the applied subgrade loading to what is considered to be a tolerable level (Figure 1.1). Nowhere in the present design method are the ballast material properties (ballast elastic modulus, etc) and
its grading considered to be significant ,parameters influencing the calculation of the required ballast depth. Consequently the following sections are primarily concerned with the calculation of the ballast depth.
Having established, for a particular sleeper type and a size, sleeper spacing such that the allowable limits of sleeper to
176
ballast contact pressure and sleeper flexural capacity are not exceeded, the next step in the conventional design procedure is to determine the required ballast depth. THE DETERMINATIONOF THE REQUIRED BALLAST DEPTH
As previously stated one of the main functions of the ballast layer is to transmit the imposed track loadings at a pressure tolerable to the subgrade. To calculate the vertical pressure on the formation caused by sleeper loadings it is essential to know
how the vertical pressure is distributed through the ballast layer. There are two main types of solutions that can be used to calculate the vertical pressure distribution with ballast depth:
. .
simplified theoretical models semiempirical and empirical solutions will be discussed of in detail in the following sections. with
These
Theoretical solutions ballast depth
the
vertical
pressure
distribution
The following are the simple theoretical solutions that have been used to calculate the vertical pressure distribution with depth. Other complex theoretical solutions have been developed specifically for road design using either multilayer theory or finite element techniques. A s this report is concerned with the current design practice of railway tracks, these solutions are outside the scope of the report. Boussinesq Elastic Theory: This theory is sometimes referred to as singlelayer ballast and the (its properties (its properties half space. The elastic theory because it assumes that the subgrade form a semiinfinite, elastic, homogeneous are constant from point to point) and isotropic are the same in each direction through a point) theory also considers the rail seat load to be
24722/8026
177
uniformly assumed
distributed area
over
a
circular the
area
equivalent the
to
the
contact
between
sleeper
and
ballast.
Boussinesq (1885) theoretically determined the stress induced at any point within a semiinfinite elastic medium by a single load Q , normal to the surface (Figure 5.1). Expressed in rectangular coordinates the following solutions were derived: (a) The vertical stress changeuZ (kPa) beneath a point load:
( J
z
= 3Q0
2T
z3
(r2 + z2 )2.5
(b) the radial shear stress change ~~(kPa beneath ) a point load:
r where
Q,
=  3Q0 = = =
rz2 2.rr (r2 + z 2 2.5
’
(5.2)
z
point load (kN), vertical depth to any point beneath the surface (m) , and
r
horizontal radius from the verticalat the position of point load to the position of any particular point beneath the surface (m) . Boussinesq equation for the of case a uniformly
Integrating
the
loaded circular area at the surface of medium the following equations, relating on the vertical axis beneath the centre been derived (Department of Scientific and
1961) :
a semiinfinite elastic to stresses at any depth of the loaded area, ha Industrial Research
(a)
The vertical stress oz(kPa) at any depth
3
Z:
Z
z
(a2 + z2 1.5
1
(5.3)
178
The horizontal stresses o x and o (kPa) at any depth z:
Y

Y
2(1 + u)z (a2 + z )0.5
Z
+
3
(a2 + z2 )1.5
The
vertical
and
horizontal
stresses
on
the
axis
are
major
and minor principal stresses respectively. The maximum shear stress rmax(kPa) at any depth z is half the difference between the principal stresses:
where
a ' a
U
= =
average uniform pressure over the loaded area (kPa), radius of the circular loaded area (m), and
= Poisson'sRatioofmaterial.
It should be noted that these expressions are independent of the modulus of elasticity, and that the vertical stress is independent of all elastic constants. This would not so be if the material had varying elastic properties. The vertical stress change in the ballastat depth can be approximated by the integration of the Boussinesq equation over the uniformly loaded area. The calculation of the minor stress and the shear stress is inaccurate due to the assumptions of a single elastic layered foundation and that this elastic medium is isotropic. In the case of sleepers the effective contact area of is a rectangular shape and consequently the solution of Equation 5.3 must
i79
I
Figure 5 . 1 Vertical stressbeneath a point load at the surface
180
therefore be modified to account for the change in the influence coefficient (i.e. that part of Equation 5.3 enclosed in brackets). While not being strictly accurate the effective support bearing area under the rail seat As, may be approximated to a circular area with a radius (m) a given by (Kurzweil 1972)
0.5
where a
L B
=
=
=
equivalent radius of the bearing area (m), effective length of sleeper support (m), and sleeper breadth (m). obtain the exact theoretical Boussinesq solutions for
In order
to
the vertical pressure at any depth caused by a uniformly loaded rectangular areaat the surface, Equation 5.1 must be double integrated over this area. For the general restated as three dimensional case Equation 5.1 can be
U
Z
=
3Q0 z3
~ T Tr
5
,
(5.7)
where
U
Z
=
= =
Q0
vertical stress at depth z beneath the load point (kPa), point load (kN),
vertical depth to any point M beneath the surface (m)l and r the distance from the origin (position of point load) to any point M (X, y r z) for which the vertical stress is sought (m) Z
z
=
(X2
+ y 2 + z2) 0.5.
181
To determine the vertical stress, oz, the summation of Boussinesq's
elementary, single, vertical concentrated load dQo on an elemen area dA = dedq over the entire rectangular bearing area A (Figure 5.2) is required, By substituting and where
U
by doZ ,
Q , by dQo
p ,
=
average uniform pressure over the loaded rectangula area, vertical stress doz, dAon because of dQo
the
infinitesimal
then follows from Equation 5.7, i.e.,
doZ
=
3pa z3 dsdrl, 5
2a r
(5.8)
and its integral is
Upon examination of this integral the closed form solution has been found to be very complex and tedius. Therefore if an a accurate estimate of the vertical stress at any depth beneath rectangular uniformly loaded area is required using the Boussine approach it is advised that the double integral, Equation 5.9 be solved using a numerical approximation technique. The ORE (1968) state that single layer elastic theory used with simplified sleeper soffit contact pressure distribution provides an adequate means of calculation of formation stresses for practical engineering purposes. In view of the degree of scatter involved in the sleeper support condition a more sophisticated approach cannot be justified. TheORE experimental data of 182
[ * U n i f o r m l y
,
Loaded
.
/ Rectangular Area
t
Note: Elementary area = d E d
Figure 5.2 Application of Boussinesq's elementary single vertical concentrated load over a uniformly loaded rectangular bearing area
183
actual subgrade pressure for various ballast depths is compared with the theoretical Boussinesq prediction of subgrade pressure in Figure5.3 (ORE 1961, Heath & Cottram 1966) . Other important findings of the ORE study are:
(a) The vertical stress distribution in the subgrade became prac tically
600
uniform at a thickness
of
construction
greater
than
mm.
(b)
Sleeper spacing in the range 630 to 790 mm had a negligible influence on the vertical stress level in the subgrade a for unit loading applied to the sleeper. Belowa Uniformly Loaded Strip of Infinite Length:
Stress
A theoretical method has been used by Eisenmann (1970) for determining the increase in vertical stress at any location under a sleeper (Figure 5.4). The sleeper in this analysisis considered
as a uniformly loaded stripof infinite length. The analysis is based upon Mohr stress circle considerations and the followin relationships have been developed: (a) The vertical stress o z (kPa) at any location (x,z):
where
p ,
e1
average uniform contact pressure between the sleeper and the ballast (kPa), and and e2 are the angles shown in Figure 5.4. (x,z):
=
(b) The horizontal stress (kPa) at any location Y
184
(c) The principal shear stress fXZ (kPa) at any location (X, z) :
T
xz

Pa 2 " i l (cos 202
 cos 2e l)
.
(5.12)
(d)
The maximum shear stress T~~~ (.kPa) :
(5.13)
Semiempirical and empirical solutions of the distribution with ballast depth
vertical
pressure
The following are the main semiempirical methods that have been developed to calculate the vertical pressure distribution with ballast. depth. Load Spread Methods: Simplified methods are often employed in practice which assume that the load is distributed vertically with a load spread slope of 1 vertical to 1 horizontal ora slope of 2 vertical to 2 horizontal. The stress distributionis also assumed uniform at any given plane velow the surface (Figure 5.5). These simplified methods calculate only the average vertical pressure at depth whereas the Boussinesq method calculates the maximum vertical pressure at a depth below the loaded area. A comparison of the vertical stress distribution calculated for both the 1:l and 2:l load spread assm2tions with the theoretical Boussinesq solution is presented in Figure 3.6; the loaded areas being circular in all cases (Department of Scientific h. Industrial Research 1961). It is clearly apparent that the assumed 2:l load spread distribution of vertical pressure more closely approximates the Boussinesq pressure distribution than the assumed 1:l load spread distribution. Clarke (1957) adopts a 1:l load spread distribution of vertical pressure through the ballast when calculating the average pressure on the subgrade. Using the Boussinesq theory to calculate the
24722/8027
185
( m m ) (in) Sleeper Soffit Level
152.5 6
305
12
457.5 18
610.0 24
762.5 30
915.0 36
0
0.5
2.5
1 .o2.0 1.5 Normalised Vertical Stress u,,~ (kPa/kN)
r
3.0
3.5
4
Legend:
" . . 
Envelope of phase l results Theoretical value (steeper spacing = 630 m m )
Phase l I Timber Sleepers H a n d (shovel)packing
@ PEH* tamping
A O n line machine tamping
Phase I I Concrete Sleepers 0 H a n d (shovel)packing 0 PEH" tamping A On line machine tamping
Figure 5 . 3 Comparison of experimental vertical stress distribution w i t h depth and the theoretical Boussinesq solution (ORE 1968)
186
Sleeper Width
Point (X, z)
X
Figure 5.4 Stress beneath auniform loadof infinite length (Eisenmann 1970b)
18 7
B
W
Uniform Contact Pressure pa
/ /
\
2
Z
\,
Figure 5 . 5 Vertical stress transmission bym e a n s o f the 2 1 distribution
188
Uniform Pressure p ,
Stress
a
2a
3a
4a
5a Variation of M a x i m u m Vertical Stress with Depth Below Circular Area (Radius a)
2:l Spread
0.2P,
v)
X
E
4a
0
3a a
2a Distance F r o m Vertical Axis 2a
Distribution of Vertical Stress o n a Horizontal Plane at Depth Under Circular Area (Radius a)
Figure 5.6 Comparison of the vertical stress distribution under a uniformly loaded circular areabased on Boussinesq equations and 1 :l and 2 : l distributions (Department of Scientific and Industrial Research 1961 )
189
maximum vertical pressure at depth beneatha uniformly loaded circular area Clarke suggests that this pressure can be estima as being between2 to 3 times the average subgrade pressure determined by the 1:l load spread distribution. For ballast depths in the range of half the sleeper breadth to twice the sleeper breadth, Clarke suggests that in order to obtain an estimate of the maximum subgrade pressure the average subgrade
5.6 and assuming that the should be doubled. Referring to Figure
circle diameter 2a approximates the sleeper breadth, it can be seen that Clark's estimate approximates the Boussinesq loaded circular area. of the maximum subgrade pressure estimate for the case a uniformly of
For a rectangularly shaped loaded area (i.e. the assumed effective area of contact beneath the rail seat) the maximum vertical pressure az(kPa) at any depth beneath the surface, can be estima as twice the average vertical pressure using1:l the load spread method. Consequently,
where
p ,
=
average uniform contact pressure between the sleeper and the ballast, depth below the surface (m), breadth of sleeper (m), and effective length of sleeper under the rail seat(m).
z
B L
=
=
=
Schramms Solution: Schram (1961) uses a method for determining the maximum vertical pressure at a depth beneath a sleeper, based upon the angle of internal friction of the ballast, which is the actual load spread of the ballast material. Schramm states that in practice the quality of the ballast,as denoted by a high angle of internal friction, has with some reservations a greater influence on the distribution of vertical pressure at depth than the length of the sleeper. The upper limit of the friction angle for coarse, rough and dry ballast is about 40°, and the lower limit for fine, smooth and moist ballast is about 30°. According
0
190
to Schralnm, the maximum vertical subgrade pressure U (kPa) for Z any ballast depth z (m) beneath the sleeper canbe calculated from
U
z
average uniform seat (kPa) contact pressure under
(5.15)
where
p ,
the
rail
sleeper length (m), distance betweenrail centres
(m) I
sleeper breadth (m) r depth of ballast layer (m) and angle of internal frictionof the ballast (degrees) Schramm also states that it is desirableto develop some vertical pressure between the sleepers thereby reducing the danger of large differences in the ballast/subgrade pressure which would result in the forcing of soft soil up between 5.7). contaminating the ballast (Figure the sleepers and
Therefore the minimum ballast depthzmin (m) required is
zmin
~
S B 2 tan8
I
(5.16)
where
S
B
e
= = =
sleeper spacing (m) sleeper breadth (m) I and angle of internal friction of the ballast.
All of the empirically derived methods of
Empirical Methods:
determining the variation of the maximum vertical pressure directly beneath the rail seat express the relationship of the vertical pressureuz (kPa) at a depth z(m) to the uniform contact pressure p , (kPa) as (Figure 5.8)
(5.17)
191
where p , is contact
defined
in al1,casesas being the entire
the
average length.
uniform
pressure
over
sleeper
Consequently, (5.18) where qr A
=
=
maximum rail seat load (kN), and entire ballast contact area of sleeper (m2). are the most commonly used empirical solutions pressure distribution with ballast depth.
The following the vertical
o
Talbot Equation: The most notable and most widely used empirical relationship is the equation recommended by the AREA and developed by Talbot (1919). The maximum vertical pressure uz(kPa) under the rail seat for any particular ballast depth is defined as
U
~
2
= = =
p a {
1
5.92 1.25’
(5.19)
where
p ,
average uniform pressure between the sleeper and the ballast (kPa) (Equation 5.18) , and ballast depth (m).
z
This equation was developed for 8’6” X 8” (2642 X 203mm) sleepers, and agrees reasonably with the observed field results except for ballast depths less than 0.1 m or greater than0.76 m. Clarke (1957) states the maximum vertical pressure uz(kPa) under the rail seatas defined by the Talbot equation can be approximated by the simplified expression 0.254)
Z
c l z
=
p , ”I
(5.20)
This assumes that the maximum intensity of pressure on the formation varies inversely with the ballast depth for any partic sleeper loading.
192
Note: Assumes that 3 adjacent sleepers resist the design wheel load therefore q, = 50% design wheel load
Figure 5.7 Maximum vertical stress on the subgrade (Schramm 1961 )
24722/8028
19 3
2
f
Subgrade
Uniform Contact Pressure Pa
Figure 5.8 M a x i m u m vertical stressdz at depth Z below the rail seat according to the empirical methods
194
Japanese National Railway Equations: The following empirical equations have been developed by the Japanese National Railway (JNR) for determining the maximum vertical pressure uZ(kPa) under the rail seat for any particular ballast depth (Okabe 1961): Horikoshi Equation:
58
0
Z
= =
P , {
1.35 10 + (1002)
(5.21)
' l
where
p ,
z
average uniform pressure between the sleeper and the ballast (kPa) (Equation 5.18) I and ballast depth for (m). Stone Ballast:
=
Okabe
Equation
Broken
uz
 P,
c
350 1.60' 240 + (1002)
(5.22)
(c) Okabe Equation for Gravel Ballast:
uz
be
P , 1
125 1.50' 50 + (1002)
(5.23)
It should
noted
that
the
above
JNR
equations
have
been
derived
for narrow gauge track conditions, where the sleeper length is 2100 mm. Therefore direct comparisons cannot be made between the Talbot equation which is relevent to standard and broad gauge track and the JNR empirical equations unless allowance is made for the difference in the average uniform pressure of the sleepers used. Comparisons of the above theoretical, semiempirical and empirical methods of determining the change in vertical pressure with increasing ballast depth for the various gauged track 5.9. The maximum vertical stress at a given presented in Figure ballast depth has been presented in the normalised form. Thereare
fore in order to obtain an estimate of the maximum vertical stress at a given ballast depth for any of the design methods
195
,/fy/ I
,
$300\///////,;4 m
,
,
.
$ . e *
/. :
400
1
. "
5.O 1 .o 2.o 3.0 4 .O Normalised Vertical Stress Oz/q, (kPa/kN)
4.0 5.0 1 .o 2.0 3.0 Normalised Vertical Stress 'Jz/q, (kPa/kN)
Narrow Gaucle Track
.. . .

Standard Gauge
Track
Note: Sleeper dimensions ( m m ) Length: 2130 Breadth: 230
Note: Sleeper dimensions ( m m ) Length: 2440 Breadth: 230
.
Normalised Vertical Stress O Z h r (kPa/kN)
Broad
Gauge
Track
Note: Sleeper dimensions ( m m ) Length: 2590 Breadth: 230 Legend:
.
Curve NO. 1
Method Simplified for case of a circular area Eisenmann Strip Load Clarke Spread Load LSocahdr a m m Spread Talbot Empirical Horikoshi Empirical Okabe (Broken stone) Empirical O k a b e (Gravel) Empirical Researcher Boussinesq
Equation N o . 5.3
5.10 5.14 5.15 5.19 5.21 5.22 5.23
Figure 5.9 Comparison of the theoretical, semiempirical,and empirical stress distributionsw i t h ballast depth
19 6
multiply the normalised vertical stress by the maximum rail seat load. In Figure 5.9 the comparison for narrow gauge track, is based on 2130 mm long X 230 mm wide sleepers, whereas the comparison for standard gauge track, is based 2440 on mm X 230 m wide
2590 mm
sleepers and the comparison for broad gauge track, is based upon X 230 mm sleepers.
Upon examinationof Figure 5.9 and selecting common ballast depths of between 200 to 300 mm it car, be seen that Schram's load spread method is a close approxixationto the Boussinesq method (assuming a circular loaded area). The Clarke load spread method appears to be a lower bound of the ballast depth requirement whereas the various empirical methods appear to be upper bounds. For large ballast depths the Talbot method closely approximates the Boussinesq solution.
(1966) have carried out static experiments to Salem and Hay determine the vertical pressure distribution with depth a €or number of sleepers with rail seat loads 89 kN of and sleeper spacings of 530 mm. Results indicate that the pressure distribution between the sleepers can be regarded as uniform for these conditions at depths of ballast greater than the sleeper spacing minus 75 mm. This observation is not entirely conclusive because the variation along the length of sleeper of the pressure distribution at depth has to be considered in any analysis.In order to attain a more uniform pressure distribution upon the subgrade not only between sleepers but along the sleeper length, greater ballast depths may be required thereby preventing excessive differential subgrade settlement and rutting of the subgrade directly beneath the sleepers.
ALLOWABLE SUBGRADE BEARING PRESSURE
As previously
mentioned one of
the
main
functions
of
the
ballast
layer is to transmit the imposed track loadinqs at a pressure tolerable to the subgrade. At present there nave been two major methods developed to determine the allowable subgrade bearing pressure:
19 l
(a) Safe Average Bearing Pressure Methods, (i) Soil Classification, and (ii) Static Load tests. (b) The British Rail Formation Design Method (Repeated Load Tests). Safe Average Bearing Pressure Methods
The ''safe average bearing pressure" of a subgrade is defined as the value of the "ultimate subgrade bearing capacity" (pressure
at which a plastic shear failure occurs) reduced by a load factor or factor of safety. The safe average bearing pressure of a subgrade is the value used in designs where the effect of set ment is considered negligible. The "allowable bearing pressure" of a subgrade is the value 'in used designs which take into account the danger of both shear failure and settlement. Con
sequently the allowable subgrade bearing pressure is the ofvalue the safe average bearing pressure that is further reduced by another safety factor, to account for possible settlement.
The values of allowable subgrade bearing pressure determined by these methods are all based upon various static testing techn applied to saturated subgrades. The assumption of the saturated subgrade condition is realistic in' that this condition is, du the open texture of the ballast, more likely to be realised under sealed roads and airfields; and it also takes into account the worse possible condition of the subgrade. The current design practice for determining the required ballast depth is to limit the subgrade pressure to some percentage, X per cent, of what is considered to be the "safe average bearing pressure" (Prause et al. 1974). This reduction factorX per cent, is adopted to prevent excessive track subgrade settlement and consequently excessive deterioration of the track geometry under service conditions.
19 8
Clarke (1957) recommends that this reduction factor be equal to 60 per cent of the safe average bearing pressurea of particular soil. The safe average bearing pressures of various soils are presented in Table5.1. Clarke also recommends as a general rule that the maximum subgrade pressure for uncompacted formations should not exceed 8 3 kPa and 139 kPa for compacted formations. The 60 per cent reduction factor used by Clarke is based upon AREA data (Talbot 1934, p. 209) which indicate that due to the variations in sleeper support and track maintenance the load on an individual sleeper could be as high as 2.7 times the nominal value computed from the beam on an elastic foundation model. Talbot also reported that the observed maximum value of the seat load frequently attained a value roughly66 per cent greater rail
than the nominal mean. Consequentlyit therefore follows that this is the main reason why Clarke has suggested the reduction factor of 60 per cent. The reduction factor of60 per cent corresponds exactly to a subgrade safety factor of 1.67. TABLE 5.1

SAFE AVERAGE BEARING PRESSURES OF SUBGRADES (CLARKE
1957)
~~ ~~~
verage ing Description Subgrade Safe Pressure (kPa)
70 75 110 145 215
Alluvial compacted ground not Made Soft clay, wet loose or sand Dry clay, firm sand, sandy clay vel Dry Compacted
 105  140  210  275
>280
The AREA (1973) recommends that the calculations of the allowable subgrade pressure be based upon laboratory tests of saturated and remoulded samples (static triaxial tests). The AREA also recommends that when estimating the bearing pressure on the subgrade the design wheel load, and therefore its corresponding maximum rail seat load, should be doubled and the calculated bearing pressure
19 9
California Bearing Ratio (C.B.R.)
2 3 4
5 6 78910
1 1 20 25 30 40 50 60 80100
I
Public Roads A D M Classification
1
I
Civil Aeronautical ADM. S o i l Classification
l
I
(kPa)Bearing Value
I
I
I
I
I
I 1 1
69
138
276 345 207
I
483
414
Modulus
27.5
of
Subgrade Reaction (lO’ kN/mm’)
I
I
54.9
41.2
220
I
137
68.7
5
California Bearing Ratio (C.B.R.)
3 4
I
1
80l00
5 6 7 8910
I
il
20 2k 30 40 50 60
I
1
1
1
1
Legend:
0  Organic W W e l l graded
G  Gravel S Sand C  Clay
M F
P
L
H
 MO.very fine sand, silt  Fines (material <0.1 m m )  Poorly graded  Low to m e d i u m  High compressibility
Figure 5 . 10 Approximate correlation o f the Casagrande, PR and CAA classification on the basis of bearing capacity (Department o f Scientific and Industrial Research 1961)
200
for this condition be compared directly with the determined safe bearing pressure. This method results in an equivalent reduction factor of 50 per cent. The California Bearing Ratio test, commonly called the CBR test, has been used extensively in pavement material and subgrade soil evaluation. It basically consists of a penetration test in which a standardised pistonof head area 3 in2 , 19.4 cm 2 , is forced into a prepared specimen (field testing can also be carried out). The load required to cause 0.1 in, 2.5 mm, penetration is compared to the load required to cause the same penetration into a high quality crushed stone which is considered to have a CBR equal to calculate the CBR of to 100. The ratio of these loads is used the material. Due to the ease of field measurement, and the good correlation between the estimated design values, the CBR method as used in road design could also be considered as being adequate for estimating the safe bearing pressure of the subgrade.
A very useful approximate correlation between the CBR value, various soil classification systems and what are considered safe
bearing pressures for compacted subgrades is presented in Figure 5.10. The CBR method and the current design method assumes that there is a unique relationship between the compressive strength & of the material and its behaviour under repeated loads (Waters Shenton 1968). Laboratory work with clay has shown that lower values of failure loads occur with repeated loading than with single load application to failure. Results published by Waters and Shenton (1969) show that in the case of clay subgrades, failure under repeated loading in a drained triaxial cell occurs at approximately 50 per cent of the rapidly applied undrained failure stress. Therefore to some extent these results confirm the AREA recommendation of using 50 a per cent reduction factor to calculate the safe average bearing pressure. a
201
British
Rail
Foundation
Design
Method
British Rail have developed a rational method of track substructure design which relates substructure requirements to the intended traffic loading on a quantitative basis (Heath, Shenton, Sparrow & Waters 1972). The developed design method is based upon extensive laboratory tests of subgrade samples and because of the
large number of samples tested the triaxial compression test w chosen. In this laboratory test two variable stresses, a major vertical principal stress ul, and two equal radial minor stresses , J ~can , be applied to the soil sample. The design method used a repeated loading triaxial apparatus which pulsed the vertical principal stressat a cycle frequency of 30 cycles per minute while maintaining the radial principal stresses at a constant confinement pressure. A square loading pulse was also chosen and the above conditions were regarded as typical of the pressure conditions experiencesby the subgrade. Preliminary tests were carried out on samples of London clay and
the resultsof these tests are shown as plots of cumulative strain versus the number of loading cycles Figure 5.11. As can be seen the results fall into two distinct groups; those in which the deformation is progressive until complete failure of the sample is reached and those in which the rate of deforma is reducing anda stable condition is being attained. Plotting elastic strain against the number of loading cycles to reach failure (defined by 10 per cent cumulative strain) a curve, similar toSN fatigue curves in metals, is obtained, Figure 5.12. On this basisa limiting repeated elastic strain can be defined, above which the deformation is continuous and below which it is terminating. Results with three different effective confining pressuresu3, indicate that the threshold stress, corresponding to this limiting elastic strain, is approximately a linear function of confining pressure. This discovered relationship forms an important part in the design method it allows as increases in the threshold stress with depth of construction. The main assumptions inherent in the proposed BR foundation design methods are (Heath al. et 1972): 202
that the threshold stress parameters for the subgrade soil may be obtained using the standard repeated loading test described that simple elastic theory can be used to in the subgrade from traffic loading that the significant traffic predict the stresses
stresses are those commonly
produced
only axle
by the static effects of the load that the water
heaviest
occurring
table at is the
topof the
subgrade.
The design method is based on the achievement of a "balanced" design whichis obtained when the ballast is sufficiently so deep that the calculated maximum deviator stress (ul U ) difference 3

induced in the subgrade by the heaviest commonly occurring axle load is equal to the average threshold principal stress difference established by laboratory repeated triaxial tests. The theoretical subgrade maximum deviator stress difference for various axle loads are shown in Figure 5.13. The threshold stress/depth curves and hence it is relationship is superimposed on these
possible to obtain the depth of ballast at which the threshold stress is equal to the stress induced by a given axle load. By taking these intersection points the design depth can be obtained. The design depth is that from sleeper bottom to the top of the subgrade indicated for the heaviest commonly occurring axle load. Using these results a design chartcan be constructed, Figure
5.14.
This design method was assessed for ballast track by performing a series of laboratory tests and track measurements. Reduced settlement rates were achieved rather consistently when the ballast depth equaled or exceeded the design depth, and the settlement rates for ballast were significantly higher. depths less than the desired depth
203
10
l
1 o2
I
1 o3
I
1 o4
I
1 os
I
I
106
1 o7
I
Number of Loading Cycles (logscale) Legend: uj u1
psi
0
,
A
12.3 11.6 1 0 . 9
u1 uj kN/m2 85
80
75 70 65 55 52.5 42.5
m
10.1
9.4 7.9 7 . 6 6.2
0
+ X
,
Figure 5 . 11 Cumulative strain resultingfrom repeated loading tests o f London clay (Heath et al 1972)
204
1 . 4 1 . 2
10% Cumulative Not Attained
(N = 106 Cycles) Strain
,z
1
1.0 S 0.8
L
m W
Threshold z .U 0 . 6  Elastic Strain  4
i0
_   
" " "
 " _ "
D
c n
0.4
? B
Y
I
0.2 O d
I
I
I
I
10
1 o2 No. of Loading
1 o3
1 o4
Strain
l(
Cycles (N)to Produce 10% Cumulative
Figure 5.12 Modified SN curve (Heath et al 1972)
205
200

400
600

800
1000
M a x i m u m Deviator Stress Difference (01
1200
0
 0 3 ) kPa for Axle Loadings Indicated
Figure 5 . 13 Derivation of the design chart relationship between induced stresses and soil strength (Heath et al 1972)
206
200
Design Axle Tonnes (kN1
400
U
8
Q
E OI
3
12 16 20 22 24
v)
; 600
r"
W
a
0
L
g
a l v)
LC
800
E
0
W
5 1000 0 m
6
E : 1200 .c
L
3
6 0 s Q d
L
E
1400
1600
1800
Threshold Stress From Standard Test (ul u 3 )kN/m2
Figure 5 . 14 Design of conventional rail track foundations (Heath et al 1972)
207
CHAPTER 6

PARAMETRIC CRITERIA
EVALUATIONOF IMPORTANT
TRACK
DESIGN
As previously outlined, current practice used in the design o rail track requires that the following design criteria are satisfied:
. .
allowable rail bending stress allowable rail deflection allowable sleeper to ballast contact pressure allowable sleeper bending stress allowable subgrade bearing pressure. to these criteria
. . .
The influence of various design parameters be discussed in datail in this section. Two standard guage mainline locomotives
currently
used
by
the
state railway systems have been selected in to order undertake the parametric analyses. The first of the two locomotives selected was the Victorian Railways (VicRail) XClass, and is illustrated in Figure6.1. This locomotive was selected because
~
it was considered representative of the current maximum axleload commonly permitted by state railway systems on mainline tracks. The second of the two locomotives selected was the Western Australian Government Railways (Westrail) LClass and is illustr in Figure 6.2. This locomotive was selected because it is representative of the axleload currently being considered by many state railway systems for future mainline freight traffic. RAIL The BENDING maximum STRESS rail AND DEFLECTION stress ESTIMATES deflection is estimated
bending
and
u
the beam on an elastic foundation analysis. The design parameters
208
c
19t
19t
c l
19t
19t1 9 t
19t
I
1
1
, 1 , n
i_z.020(zozO
j
10.060 B o l s t e r Centres 18.360Pullina Lines
,
i
L U
I.
W
2.020
2.020
1
Figure 6 . 1 Victorian Railways main line diesel electric X class locomotive
24722/8030
209
U
l
U
U
IJ
k
/ l *
4
22.9t 22.9t 22.9t
i
12.500 Bolster Centres 20.220Pulling Lines
4
,905L 'l : .905J
*
Figure 6 . 2 W e s t Australian Government Railways main line diesel electric L class locomotive
210
that influence and deflection
the magnitude of the include:
maximum
rail
bending
stress
. . . .
.
the rail section used in the design the assumed valueof the track modulus per rail the axle load of the vehicle the axle spacings of the vehicle the speed of the vehicle.
The design vehicle combinations used in the parametric analysis of the rail bending stress and deflection were two coupled Vicrail XClass locomotives and two coupled Westrail LClass locomotives. The above vehicle combinations were selected in order to obtain the maximum rail bending stress and deflection caused by the interaction of adjacent axle loads. Three vehicle speeds were selected, viz. 80, 100 and 120 km/h and the variation of the maximum rail bending stress and deflection with changes in the assumed track moduli values were plotted for each of the two above vehicle combinations and for 47 and 53 kg/m rail sections. These plots are presented in Figures 6.3, 6.4, 6.5 and 6.6. These Figures also outline the A.R.E.A. limits of the maximum allowable rail bending AND stress and deflection. ESTIMATES
RAIL.BENDING
STRESS
DEFLECTION
The maximum allowable rail bending stress for continuously welded rail, as determined by the simpleA.R.E.A. factor of safety method, was found to equal138 MPa, (Equation 3.38) and estimating the yield strength of standard carbon rail steel to equal 410 MPa. The maximum allowable rail deflection as recommended by the A.R.E.A. is equal to6.35 mm. Westrail have published values of the modulus tack per rail for
standard gauge mainline track, (see Table 3.5).
211
From this table
"
2 H
41
:
1 80
 170
160
Two Westrail
Lclass Locomotives
. . .. I20 .
,
I 150
140
C .
A.R.E.A. Allowable ,Rail Bending Stress

'..
?
m
130
$ 120 a
5 E 'Z
H
110 100
90
70i ' I
10
Typical State Railway Track Modulus Value
15
20 25 30 Track Modulus Per Rail ( M P A )
35
40
Figure 6 . 3 Variation o f the m a x i m u m rail bending stress l i m i t assumed track moduli values for variouslocomotives at speed, using 47(kg/m) continuously welded rails
212
160 150 140
3 /A.R.E.A. Allowable Rail Bending Stress Limit
’.
130 2
..
E
._
H
? 4
60 70
Xclass Locomotives
50
40
10
‘1
Typical State Railway Value Track Modulus
l
I
I
15
20 Track
25 35 30 Modulus Per Rail ( M P A )
40
Figure 6.4 Variation of the m a x i m u m r a i l bending stress l i m i t with assumed track moduli values for various locomotives at speed, using 53(kg/m) continuously welded rails
213
0 .c
c
8.0
E
.5
7.0
r x
W
. A.R.E.A.R e c o m m e n d e d
Deflection Limit
T w o Westrail Lclass locomotives
3.0
2.0
1 . 0
'Typical
Track Modulus Value
U
State Railway
Two Vicrail
Xclass locomotives
I
;
1 0
I
1
40
15
25 30 Track Modulus Per Rail (MPA)
20
35
Figure 6 . 5 Variation of the maximum rail deflection w i t h assumed moduli for variouslocomotives at speed using 47(kg/m) rails
214
Track Modulus Per Rail (MPA)
Figure 6 . 6 Variation ofthe m a x i m u m rail deflection w i t h assumed track moduli for variouslocomotives at speed, using 53(kg/m) rails
215
it is apparent that an estimate of the track modulus per rail 15 MPa is typical of standard guage timber sleepered mainline track of normal construction. It is apparent in Figure 6.3 than when 47kg/m rail sections are used in conjunction with the typical track modulus estimate, the combinations of two coupled Westrail LClass locomotives exceed the allowable rail bending stress criterion for speeds greater than 80 km/h. In fact for speeds of120 km/h this rail design criterion is not satisfied unless the assumed track modules per rail is estimated as 33
M
By comparison with Figure 6.4 it is clearly evident that the 53 kg/m rail section in conjunction with the typical track modulus estimate is adequate for all the locomotive combinations and speeds considered.
Examining Figure6.5 and 6.6 it is apparent that when 47 and 53 kg/m rail sections are used in conjunction with the typical tra modulus estimate that the combination of two coupled Westrail LClass locomotives exceeds the allowable rail deflection criter for speeds greater than 80 km/h. For speeds of120 km/h the rail deflection criterion is satisfied if the assumed track modulus per rail for both 47 and 53 kg/m rail sections is estimated as being greater than 17.5 MPa.
The significance of the assumed track modulus per rail upon t resulting maximum rail bending stress and deflection estimates is ,
summarised in Table 6.1 for the caseof a 47 kg/m rail section. It is readily apparent that the assumed track modulus value is more influential upon the value of the maximum rail deflection than the maximum rail bending stress. If the value of the track 100 per cent from 10 to 20 MPa the maximum modulus were varied rail deflection is reduced 49 by per cent compared with a 10 per cent reduction in the maximum rail bending stress.
Adopting the typical track modulus value of 15 MPa, estimates o the maximum rail bending stress and deflection were calculated for the two design vehicle combinations at various speeds and for the 47, 50, 53 and 60 kg/m rail sections. These estimates are 216
TABLE 6.1
 EFFECTS THAT THE VARIATION OF TRACK MODULUS HAS UN THE MAXIMUM RAIL
VARIOUS ___ LOCOMOTIVES A T 100 km/h AND USING 47 kg/m RBILS Maximum Rail Bending Increase Stress at 100 km/h in Track cent) (wa) Kodulus (Per cent) Two VicRail Two Westrail XC lass LClass locomotives locomotives Reduction in Maximum Rail Bending Stress (Per
Two Vicrail XClass locomotives
BENDING S T R E S S .  ~ ~ D ~ E F L E C ~ FOR ION
Track
Modulus
4 p
Maximum Rail Deflection at 100 km/h (m) XClass LClass locomotives Locomotives
8.0 4.1 2.9
2.3
Reduction in maximum Rail Deflection (Per cent)
~ ~ ~~
per Rail.
Two Wcstrclil Two VicRail
Two Wcstrail Two VicRail
Two Westrail
10
20 30
+loo
e200
+300

132.6
156.6


LCLass locomotives
XClass LClass Locomotives locomotives
9 16 21
“_l____
10.2
19
49
65 73
40
119.0 110.5 105.6
142.3 132.2 124.0
10 17 20
5.2
3.6
2.8
G4 71
presented in detail in Table 6.2. Using this data as a basis the influence on the maximum rail bending stress and rail deflection to variationsin the design speed, vehicle combination and rail section were appraised. The effect of increasing the speed and the axleload of the design vehicle combination upon the maximum rail bending stress and deflection is summarized in 6.3. Table It was found that all the rail sections considered gave very similar results to the percentage variation. Although the
absolute magnitude of the maximum rail bending stress and defl were naturally at variance. The important aspect illustrated in Table 5 is that a 50 per cent increase in the speed, i.e. 80 to 120 km/h for both vehicle combinations resulted in an increase the maximum rail bending stress and de€lection 15 per of cent in both cases. Whereas increasing the axleload by 20 per cent, i.e. replacing the XClass locomotives with LClass locomotives, at identical speeds resulted in an increase in the maximum rail bending stress and deflection of 18 and 27 per cent respectively. It should however be emphasised that part of the increase in rail bending stress and with deflection is due the higher the to the altered axle spacings in conjunction The effect of axle loads. size upon the
in
increasing
rail
section
maximum
rail bending stress and deflection is summarised in 6.4. Table When the typical value of the track modulus is used in conjun with the rail sections considered, it was discovered that increasing the rail size alone only influenced the rail bending stress to any marked degree. This is illustrated by the effect of increasing the rail sectional strength by 85 per cent, i.e. increasing the rail section from 47 to 60 kg/m, is to reduce the maximum rail bending stress by31 per cent, but only reduces the maximum rail deflection by less than0.02 per cent. , This further reinforces the observation made in Table 6.1 that the maximum rail deflection is strongly dependent upon the assumed track modulus value.
218
TABLE 6.2
 ESTIMATES OF
THE MAXIMUM RAIL BENDING
STEESS
AND
DEFLECTION AND
AND AXLELOADS CALCULATED FOR VARIOUS RAIL SECTIONS, SPEEDS USING AN ASSUMED VALUE OF THE TRACK MODULUS OF 15 MPa
Rail Maximum Rail Maximum Bending Stress (MPa) Deflection Rail Section lCVClass LClass XClass (kg/m) 47 7.3 50 5.7 Speed
(km/H1
(mm) lass
2 VicRail
2 Westrail
VicRail 2 Westrail 2
locomotives locomotives locomotives locomotives 138.0 80 100 148.1 120 158.1 80 100 120
100 120 60
117.1 125.6 134.1 104.9 112.5 120.2 123.4 132.3 6.8 141.3 7.3 108.7 6.3 6.8 124.5 7.2 94.8 108.5
5.0 5.4
5.0 5.3 5.7
6.4 6.8 6.3
80
53 116.6
92.2 98.9 105.6
80.1 85.9 5.6 91.7
4.9 5.3 5.7
4.9 5.3
80 LOO 101.7 L20 7.2
6.3 6.8
TABLE 6.3
 EFFECT OF
INCREASING SPEED AND AXLELOAD ON THE MAXIMUM RAIL BENDING STRESS AND DEFLECTION FOR THE RANGE OFRAIL SECTIONS CONSIDERED AND USING AN ASSUMED TRACK MODULUS VALUE OF 15 PMa (PER CENT)
h ) h)
0
For all rail section Increasing speed for idential Increasing axleload from considered Xclass to LClass for Speed range 80 to Speed range 80 to identical speeds (20 per 100 km/h 120 km/h axleload centincrease) (25 per cent speed (50 per cent speed increase) increase) Increase in maximum rail bending stress Increase in maximum rail deflection
7
15
18
7
15
27
Increase Increase
of rail rail Reduction
of
Rail moment cent) Section (Per
sectional sectional weight second speed of
in .the maximum rail bending stress at a constant of 100 km/h cent) (Per
Reduction in the maximum r a i l deflection at a constant speed of 100 km/h (Per cent)
1.25.6 MPa
1 4 R. 1. MPa
11 21 0.01 31
so
53
60
l " " . " . _ . .
6
26
H5
10
21 32
5.4 m m 0.01
6.n m m
0.Ul
13
28
~
0.01 0.02
0.01

"~
SLEEPER TO BALLAST
CONTACTPRESSURE ESTIMATES
Before the sleeper to ballast contact pressure can be estima the sleeper rail seat load must be evaluated. Two methods of determining the rail seat load, the beam on an elastic found method and theA.R.E.A. method, (Equations 4.6 and 4.9) were evaluated for timber sleepered track for the two design vehicl combinations outlined in the previous section. The variation of the maximum rail seat load with sleeper spacing for the two design vehicle combinations and using the two alternate rail load formulae is presented in Figure6.7 and 6.8. Under existing state railway conditions the typical sleeper spacing of standar guage mainline timber sleepers is commonly 610 mm. Using this sleeper spacing in conjunction with Figure 6.7 and 6.8 it is
evident that the maximum rail seat load, as determined by the beam on anelastic foundation method, is nearly constant for a rail sections considered and for assumed track moduli values the range of 10 to 20 MPa. The effect of varying the rail section and track moduli on the maximum rail seat load for a sleeper spacing of 610 mm and the beam on an elastic foundation
analysis is evaluated for the two vehicle combinations travelli at speeds of 100 km/h. The results of this evaluation for the two coupled Vicrail XClass locomotives and the two coupled Westrail LClass respectively. locomotives are presented in Tables 6.5 and 6.6
From a comparison of Tables 6.5 and 6.6 it is evident that 47 to 60 kg/m, i.e., an 85 per increasing the rail section from cent increase in rail second moment of inertia, results in a decrease of the estimated rail seat load of between 1 and 2 per cent using the beam on an elastic foundation method and assumed track moduli values of between 10 and 20 MPa respectively. The effect of increasing the track moduli values from 40 10 to MPa results in an increase of the rail seat load of between 11 and 15 per cent for47 kg/m rail sections and between 4 and 5 per cent for 60 kg/m rail sectioQs. From the above observations it is clearly apparent that for tracks of normal
222
construction
and
75
70
65
Z
60
! 2
Y
S
+
55
. m
v)
m al
[ r
50
45
40
1
I
35
I
500
600
I
6lOrnm SleeperSpacing 700
800
Sleeper Spacing (mm) Legend: B.H.P. Rail Sections 47 kg/m 50 kg/m 53 kg/m 60 kg/m

"
Note: Speed 100 km/hr A.R.E.A.Impact Factor = 1.51
k
= Assumed Track Modulus per Rail k ( M P a )
Figure 6.7 Estimates o f the variation o f rail seat load with sleeper spacing for Vicrail Xclass locomotives
223
100
95
/
90
85
z
Y
I + m m
80
;
._
75
2 a
70
65
60
55
50 500
600
700 Sleeper Spacing (mm)
Legend: B.H.P.Rail Sections 47 kglm . 50 kg/m 53 kglm 60 kglm
" " "
"
Note: Speed 100 kmlhr A.R.E.A. Impact Factor = 1.51 k = Assumed Track Modulus per Rail k (MPa)
Figure 6 . 8 Estimates of the variation of rail seat loadw i t h sleeper spacing for Westrail Lclass locomotives
224
e
N N
N
4
m
,
TABLE 6.5
0
W
 EFFECT OF
VARYING RAIL SECTION AND TRACK MODULI ON TIIE BEAM ON AN ELASTIC FOUNDATION ESTIMATEOF THE RAIL SEAT LOAD, FOR VICRAIL XCLASS LOCOMOTIVES AND SLEEPER SPACING OF 610 I N l l
Track R e m on an elastic foundation Effect on rail seat load of 47 to modulus per estimate of rail seat load increasing rail from rail (kN) 60 kg/m (MPa)
4 7 Rail cent) 60 (Per Rail Section (kg/m) Section (kg/m) 48.7
50.2 56.0
_ I _ _ _
Effect on rail seat load the track moduli values (Per cent)
47 Section Rail (kg/m)
of increasing
" _ _ _ " _ _ I _ _ _
60 Section Rail
( k d m)
10
20
40
48.2 49.0
50. R
_ _ _ " ^ I _ 
1 2
3
2
9
15
_"
. . ._
. .
5
"
TABLlC 6 ,6
 EFFECT
ON VARYING IlAlL Sh''g.TJlNlD.TRACK MODULI ~CINl_TIIE UEAM ON A N J3JASTIC I'OUNllATION SEAT LOAD, FOR WESTRAIL LCLASS LSOMOTIVES A N D A SLEEI'ER SPACIN(: OF 610 m m
_ " I _ _ 
_ I I _ _ _ _ _
ISSTIMATP: OF 'HW RAIl!,
_ . _ _ "
____.

rail
Track Beam on an elastic foundation modulus per estimate of rail seat load per (kN1
(MPa) 47 Rail Section 6 0 Rail Section (kg/m) (kg/m) (ky/m)
62.3 61.5
Effect on rail seat loadof Effect on rail seat loadof increasing rail section from increasing tack moduli values 47 (Per to 60 kg/m cent)

_ "
(Per cent)
47 Rail Section
60 Rail Section
( k d m)
10 20
40
1.
2
+2
+2 +4
I_,.
_ _ I
63.8
69.0
62.5 64.0
7
_____._
+l1
__
" "

with
assumed
track
moduli
v,alues
of
between 10 and 20 MPa
the
beam on an elastic foundation prediction of the rail seat load can be regarded as being nearly constant for all rail sections and only dependent upon the sleeper spacing. Due to this evident lack of influence on the rail seat load of both the rail sectio size and track moduli value in the ranges considered, it was decided to base the parametric study of the sleeper to ballast interaction on the A.R.E.A. The A.R.E.A. method of design approach. the rail seat load results
determining
in
an estimate which is only dependent upon the sleeper spacing; rail section size is considered to have no significant influen Upon closer examination of Figure 6.7 and 6.8 it is also evident that the slope of the beam an on elastic foundation estimates of the rai~l seat load for both locomotives on track with a moduli of between 10 and 20 (MPa) closely approximated the slope of the A.R.E.A. rail seat load estimate.In Chapter 2 it was noted in rail seatload the discussion of the development of A.R.E.A. the estimate thatit was based upon the beam as an elastic foundation estimate usinga 57 kg/m rail section,a track modulus per rail of 13.8 MPa and a configuration of four 150 kN wheel loads spaced
at 1.800, 2.000 and 1.800 m centres. Since the influence of the rail section size and typical tack moduli values has been foun to be negligible upon the beam on an elastic foundation estimate of the of the rail seat load it is considered that the slope estimate is mainly governed by the axle spacings of the design vehicle group.
As already mentioned above, the parametric study of the sleeper to ballast contact pressure was undertaken using A.R.E.A. the sleeper design method. The variationof the maximum sleeper to ballast contactpressure'with sleeper spacing and for various
,
sleeper
lengths
for
the
design
vehicles
travelling at 100 km/h is
presented in Figure6.9.
The recommended limit of the sleeper to
ballast contact pressure, 450 kPa is also indicated in the diagrams. Clearly the A.R.E.A. recommended limit of the allowable sleeper to ballast contact pressure
226
for
timber
sleepers
is
400
II
300
I
610mm Sleeper Spacing
600 700 Sleeper Spacing ( m m )
Note: Design Conditions Standard gauge track Speed 100 km/h A.R.E.A. Impact Factor A.R.E.A.Sleeper Design Method
z
0
Other Sleeper Dimensions Breadth: 0.230m Thickness: No influence o n contact pressure
Figure 6 . 9 Variation o f the sleeper to ballast contact pressure w i t h sleeper spacing
227
exceeded by nearlyall
the
sleeper
lengths
considered
under
the
Westrail LClass locomotive loading at 100 km/h. This recommended limit is also exceeded under the Vicrail XClass locomotive
loading even for sleepers with surface dimensions of 2.440 X 0.230 m spaced 610 mm apart, (i.e., the typical track configuration of the majority of standard guage mainline timbered sle track used by the state railway systems). Consequently, the following two general conclusions can be made, viz. either the existing tracks are underdesigned and require larger sleeper dimensions and closer sleeper spacings for the axleload and speed conditions; or, that the A.R.E.A. recommended design criterion of the allowable sleeper to ballast contact pressure has been empirically derived for a lighter axle load or smaller sleeper spacing. The track configuration mentioned above has been the standard guage track standard by used the majority of state railway systems for many years, it and is most probable
that this standard was developed for a lighter axleload. Even though tracks with this configuration have exceeded this conta pressure criterion, they have performed adequately for the past twenty years under axleloadings similar to that of the Vicrail XClass locomotiveand, more recently under axleloading similar to the Westrail locomotive, (although probably at operating speeds of less than 100 km/h). This fact raises the question of the validity of this contact pressure criterion.It is therefore considered that until the allowable sleeper to ballast contact pressure criterion for specific axle load conditions are related to the standard of track maintenance, no reliable estimate of actual contact pressure safety factor, which is inherent in the current A.R.E.A. track design procedure, can be made. The effect of varying the sleeper length upon the sleeper to ballast contact pressure is outlined in Table 6.7 for the commonly used sleeper spacing of 610 m m . As expected the reduction in the contact pressure is linearly proportional to the increase in the sleeper to ballast surface contact area, (Equation 4.17). The values of the sleeper to ballast contact pressure are quoted in
228
TABLE 6.7
 EFFECT
OF VARYING SLEEPER LENGTH AND THE DESIGN LOADING CONDITION THE UPON A.R.E.A. ESTIMATE OF THE SLEEPER TO BALLAST CONTACT PRESSURE, FOR A SLEEPER SPACING OF 610 mm
" "
Sleeper Change in Sleeper to ballast contact Change in sleeper Length* Sleeper pressure estimated by the contact pressure (m) Length cent) A.R.E.A. (Per method design (KPa)
N N ID
to ballast
(Per cent) VicRail Westrail increase to DueDue XClass
2.440 554 514 470
increase to
LClass
of sleeper length
of axleload
I
534
+L8
+l8
+l8
" "
452
2.540 2.640
+4
4
8
+8
436
Note:
Other sleeper dimensions are breadth 0.230 m, and thickness (irrelevant to results).
Table 6.7 in variations SLEEPER
order
to
illustrate and
to
the
designer
the
influence of
of
axleload STRESS
sleeper
length.
BENDING
ESTIMATES
The two preselected design vehicles; viz. the Vicrail XClass and the Westrail LClass locomotives, travelling at speeds of100 km/h were used to evaluate the variation of the sleeper bending stress with sleeper spacing.The variation of the sleeper bending stress with sleeper spacing was also investigated for a range of sleeper lengths and thicknesses, the reference sleeper dimensions being in both instances 2.440 X 0.230 X 0.115 m. It was considered that, due to the form of the design equations f determining the sleeper bending stress, the effect of varying the sleeper length and thickness would be more dominant than
that
the sleeper breadth. In the following analyses the A.R.E.A. method calculating the rail seat load and the Schramm method calculating the sleeper effective length were used. The sleepers were assumed to be fitted with rail bearing plates in the anal Effect of varying bending stress the sleeper length upon the maximum sleeper
Two methods of calculating the maximum sleeper bending at moment the rail seat were evaluated, there being Battelle/Schrarrm the method and the A.R.E.A. method (Equations 4.21 and 4.38).
Sleeper lengths of 2.440, 2.540 and 2.640 m were used in the analysis to ascertain the influence of varying the sleeper len alone upon the sleeper bending stress at the rail. While the sleeper length was varied the other reference sleeper dimensio
~
were
held
constant.
Using the Battelle/Schramm bending stress method the variation the sleeper bending stress at the rail seat with the sleeper spacing for the range of sleeper lengths and axleloads considered is presented in Figure 6.10. 230
I
~~
500
600 Sleeper
700
Spacing ( m mj
0 8l 
Note: Design Conditions Other Sleeper Dimensions Standard gauge track Breadth: 0.230 m Speed 100 km/h Thickness: 0.1 15 m A.R.E.A. Impact Factor BatteIle/Schrarnm Bending Stress Method Sleeper plates fitted to sleeper
Figure 6 . 1 0 Variation of the sleeper bending stress at the r a i l seat w i t h sleeper spacing using theBattelle/Schramm method. (For variable sleeper lengths)
231
It is clearly apparent that the A.R.E.A. recommended design limit of 7.6 MPa, is exceeded for all the sleeper dimensions and axleloads considered. For sleeper spacing of610 mm, commonly used by the state railway systems the effect of varying the sleeper length upon the sleeper bending stress at the rail seat is presented in Table 6.8. It can be seen that at this sleeper spacing an 8 per cent increase in the sleeper length causes a 33 per cent increase in the sleeper bending at stress the rail seat, whereas increasing the axleload from that of a Westrail LClass locomotive between 18 to 20 per cent. The above analysis was repeated using A.R.E.A. the bending stress interest that of a Vicrail XClass causes an increase of to
method and the results were presented in Figure 6.11. Of
is the observation that when this method is used in conjunct with the Westrail LClass locomotive 2.540 the m longer sleeper exceeds theA.R.E.A. recommended bending stress limit at a sleeper spacing of 560 mm. The variation of the sleeper bending stress at the rail seat with the sleeper length a for sleeper spacing of 160 mm is presented in Table 6.9. It is evident that increasing the sleeper length by 8 per cent in conjunction with the A.R.E.A. bending stress method causes a 62 per cent increase
in the sleeper bending stress at the rail seat. The effect of increasing the axleload from that a of Vicrail XClass to that of a Westrail LClass locomotive increases the sleeper bending stress at the rail seat by between 19 and 22 per cent; which is similar to the Battelle/Schraqm method result. Of interest is the observation that although the 8 per cent increase in sleeper length resultsin a 62 per cent increase in the bending stress a the rail seat for the A.R.E.A. method, the maximum level of stress predicted is still roughly 60 per cent of that predicte by the Battelle/Schramm method. The Battelle/Schramm method of estimating the maximum sleeper bending stress at the rail seat could therefore be regarded as than that of the A.R.E.A. method. being a more conservative estimate
232
N
J .
.
N m
W
0
P N
c
TABLE
6.8. EFFECT OF VARYING THE SLEEPER LENGTH UPON THE 'SLEEPER BENDINGATSTRESS THE RAIL SEAT CALCULATED BY THE SCHRAMM/BATTELLE METHOD, FOR A SLEEPER SPACINGOF 610 mm
"
Sleeper Change in Length* Sleeper Length (m) Schramm/Battelle the method+
h ,
Sleeper bending stress the estimated cent) seat (Per rail by (MPa)
at
Change in sleeper bending stress
"
W
W
(Per cent) VicRail Westrail increase to Due Due sleeper XClass length LClass of 2.440 2.540 13.4 2.640
increase to of axleload
+4 +8
" 
10.0
12.0
+l6 +33
+20
+l9
11.6
13.8 15.8
+l 8
TABLE 6.9
 EFFECT OF
VARYING THE SLEEPER LENGTH UPON THE SLEEPER BENDING STRESS THE RAIL AT SEAT CALCULATED BY THE A.R.E.A. mTHOD, FOR A SLEEPER SPACING 'OF 610 mm
"
"
Sleeper Change in Sleeper bending stress at the rail Length* Sleeper seat estimated the by A.R.E.A. method (m) Length (MPa 1 (Per Cent) VicRailXClassWestrailLClass 2.440 2.540 2.640 6.2
+
Changes in Sleeper bending stress (Per cent)
"
Due toincrease Due toincrease of sleeper length of axleload
+4
+8
5.2 6.7
8.3
8.1 10.1 +30
" 
+l9 +21 +22 ~_ ""
+62
Note:
* Other sleeper dimensions are breadth 0.230 m, thickness 0.115 m. + Sleeper bearing plates fitted.
13
Sleeper Length
1
'r"
I
I I I
I
3
I L
610 m m Sleeper Spacing
I
I
800
500
600 700 Sleeper Spacing ( m m )
Other Sleeper Dimensions Breadth: 0.230 m Thickness: 0.115 m
Note: Design Conditions Standard gauge track Spe 1 ke 0 md 0 /h A.R.E.A.Impact Factor A.R.E.A.Bending Stress Method Sleeper plates fitted to sleeper
Figure 6 . 1 1 Variation of the sleeper bending stress atthe rail seat w i t h sleeper spacing using the A.R.E.A. method. (For variable sleeper lengths)
235
In Chapter4 it was suggested that the maximum bending moment at the centre of the sleeper. should be calculated on the basisof an assumed uniform bearing pressure over the overall sleeper length (Equations 4.23 and 4.39). Using this method the variation of the sleeper bending stress at the centreof the sleeper with sleeper spacing for the range of sleeper lengths and axleloads considered is presented in Figure 6.12. It is clearly apparent that theA.R.E.A. recommended sleeper bending stress limit of 7.6 MPa is exceeded by all the sleeper sizes considered. The variation of the sleeper bending stress at the centre of the sleeper with sleeper length for a sleeper spacingof 610 mm is presented in Table 6.10. It is evident thata 8 per'cent increase in the sleeper length results in a 36 per cent reduction in the maximum
bending stress at the centre of the sleeper. Whereas increasing of a Vicrail XClassto that of a Westrail the axleload from that LClass increases the sleeper bending stress at the centre of the sleeper by between 17 and 21 per cent. It can generally be stated that increasing the sleeper length results in an increas of the sleeper bending stress at the rail seat and a reduction of the sleeper bending stress at the centre of the sleeper. Effect of varying bending stress The above the sleeper thickness upon the maximum sleeper
comparison
of
the
Battelle/Schramm A.R.E.A. and methods
of determining the bending stess at the rail seat was repeated but for this case the influence of varying the sleeper thickness alone was investigated. The three sleeper thicknesses slected
were 0.115, 0.135 and 0.155 m and the other reference sleeper dimensions were held constant. Using the Batelle/Schramm bending stress method the variation of the sleeper bending stress at the rail seat with the sleeper spacing, for the range of sleeper thicknesses and axeloads considered, is presented in Figure 6.13. It is significant that the A.R.E.A. criterion of the maximum allowable sleeper bending stress is only satisfied for relativel large sleeper thicknesses. For the commonly used sleeper spacing of 610 mm, the effectof varying the sleeper thickness upon the
236
TABLE 
6.10
 EFFECT
" "
OF VARYING
"
THE SLEEPER LE:NGTH UPON THE SLEEPER BENDINGSTRESS AT THE SPACING OF 610 mm
" "
____
S
CENTRE FOR A SLEEPER Sleeper Change in Length S leeper (m) length
h ,

Sleeper bending stress at centre Change in sleeper bending stres W") cent) (Per
____ " _
" Due to an increase in axleload
VicRail XClass Westrail
W l.
LClasS
Due to an increasc of sleeper lcngth
2.440 2.540
"4
+8
"
1.8.6
3.5.2
22.5
19 36
" 
f2 1
+1.9
18.1
2.650
12. l
14.2
"  ~ 
+l7
Note: Other sleeper dimensions are breadth
0.230 m, thickness 0.1.15 m.
8EZ
6EZ
sleeper bending stress at the rail seatis presented in Table 6.11. It can be seen that at this sleeper spacing 35 a per cent increase in the sleeper thickness causes 45 a per cent reduction in the sleeper bending stress; whereas the increase in axleload from thatof a Vicrail XClass to that of a Westrail LClass results in an18 per cent increase of the sleeper bending stress. Similarly for the A.R.E.A. bending stress method, the variation
rail seat with the sleeper spacing, for the range of sleeper thickness and axleloads considered, is presented in Figure6.14. It is clearly evident that the sleeper sizes considered satisfy A.R.E.A. the criterion of the maximum allowable sleeper bending stress for the range sleeper spacings that would be used in track design. For the commonly used sleeper spacing of 610 mm, the effect of varying the sleeper thickness upon the sleeper bending stress at the rail seat is presented in Table 6.12. As with the Battelle/Schramm method detailed above the effect of increasing the sleeper thickness by 35 per cent in conjunction with A.R.E.A. the method results in a 46 per cent reductionof the sleeper bending stress at the rail seat. What is significant is that the sleeper, 50 bending stresses calculated by the A.R.E.A. method are roughly per centof those calculated by the Battelle/Schramm method. The variation of the sleeper bending stress at the centreof the sleeper with the sleeper spacing, for the range of sleeper thicknesses and axeloads considered, is presented in Figure 6.15. As observed in Figure 6.12 the A.R.E.A. recommended bending !d. stress limitof 7.6 MPa is exceeded by all sleeper sizes considere For the commonly used sleeper spacing of 610 mm, the effectof varying the sleeper thickness upon the sleeper bending stress at the centre of the sleeper is presented in Table 6.13. As to be expected, the35 per cent increase' in the sleeper thickness reduces the sleeper bending stress at the centreof the sleeper by 45 per cent. The maximum sleeper bending stress at the centre of
of the sleeper bending stress at the
of the
sleeper
for
the
centrebound
240
condition
is
approximately
I w
e
TABLE
6.11
 EFFECT OF
RAIL 610 mm
VARYING THE SLEEPER THICKNESS UPON THE SLEEPER BENDING STRESS AT= SEAT CALCULATED BY THE SCHRAMM/BATTELLE METHOD, FOR A SLEEPER SPACINGSE'
N P
Sleeper Change in Sleeper bending stress at rail seat Thickness*sleeperestimatedbytheSchramm/Battellemethodstress thickness (m) (Per cent) VicRail XClass Westrail LClass Due increase 0.115 0.135 0.155
+
Change in sleeper bending
"
to increase of sleeper thickness load axle
Due b
of
+l7
10.1
6.6 +35
7.4 5.6
11.9 8.6
27
45
+l8
+l8
+l 8
"
Note:
* Other sleeper dimensions: length 2.440 m, breadth 0.230 m + Sleeper bearing plates fitted
TABLE 6.12
 EFFECT OF
in
VARYING
THE
SLEEPER
THICKNESS
UPON
THE
SLEEPER
BENDING STRESS AT THE
RAIL SEAT CALCULATED BY THE A.R.E.A. METHOD, Sleeper Change Thickness* sleeper (m)
h , h ,
A SLEEPER FOR
SPACINGOF 610 mm
"
Sleeper bending stress at the rail Change in sleeper bending stress seat estimated by the AREA method+ (Per cent) (MPa) Westrail Due to increase Due
thickness (Per cent) VicRail axleload XClass LClass sleeper thickness in
6.3

"
to increase
0.115
+l7
+35
5.2 3.7
2.8
4.5
3.4 29 46
_"
+21 +21 +21
"
0.135 0.155
Note:
* Other sleeper dimensions: length 2.440 m, breadth 0.230 m + Sleeper bearing plates fitted.
TABLE 6.13
 EFFECT
CENTRE
OF VARYING THE SLEEPER THICKNESS FOR A SLEEPER 'SPACING 610 OF mm
UPON
THE
SLEEPER
BENDING
STRESS
AT
THE
" "
sleeper
N
B P W
Sleeper Change in Sleeper bending stress at centre Change in sleeper bending stres thickness (m) thickness VicRail cent) (Per Westrail an to Due Due an to las XC s increase LClass of in increase sleeper
0.115
S
" "
+l 7
18.6
0.135 0.155
+35
13.5 10.2
22.5 16.3
12.4
27 45
+21 +21
+21
"
Note:
Other sleeper dimensions: length
2.440 m, breadth 0.230 m.
I
I.610mm
Isleeper Spacing 1 , 500
I
Jl
700 600 Sleeper Spacing ( m m ) Other Sleeper Dimensions Length: 2.440m Breadth: 0.230m
800
Note: Design Conditions Standard gauge track Speed 100 k m / h A.R.E.A. Impact Factor A.R.E.A. Bending Stress Formula Sleeper plates fitted to sleeper
Figure 6 . 14 Variation of the sleeper bending stress at the rail seat w i t h sleeper spacing, using the A.R.E.A. method. (For variable sleeper thickness)
244
"
L
c
c
c
c
a
500
i
600 700 Sleeper Spacing (mm)
aoo
I
Note: Design Conditions Standard gauge track Speed 100 k m / h A.R.E.A. Impact Factor
Other Sleeper Dimensions Length 2.440 m Breadth 0.230m
Figure 6 . 15 Variation of the sleeper bending stress at the centre with sleeper spacing. (Forvariable sleeper thickness)
245
50
to
100
per
cent
greater
than
the
maximum
sleeper
bending
stress calculatedat the rail seat. It can clearly be seen in Figures 6.12 and 6.15 thesleeper bending stress at the centreof the sleeper also exceeds the A.R.E.A. recommended sleeper bending stress limit of 7.6 MPa for the range of sleeper lengths and thickness considered. It was decided to determine what the
required sleeper dimensions would have to be in order that th sleeper bending stress at the centre would satisfy or approach
the A.R.E.A. design limit. In the parametric analyses above the reference sleeper dimensions used are 2.440 X 0.230 X 0.115 m. Referring to Figure 6.15 it can be seen that the maximum sleeper bending stress at the centre of the sleeper for a sleeper spac of 610 mm is 18.5 MPa for the Vicrail Xclass locomotive, whereas for the Westrail LClass locomotive this stress22.5 is MPa. per Noting in Table 6.12 that increasing the sleeper length8 by cent, viz from 2.440 to 2.640 m, that the reduction in the bending stress at the sleeper centre is 36 per cent, and that in Table 6.13 that increasing the sleeper thickness 35 byper cent, viz from 0.115 to 0.155 m, and the reduction in the bending stress at the sleeper centre is 45 per cent, the following estimate of the sleeper stress at the centre 2.640 of a X 0.230 0.155 m sleeper can be made: (a) For the Vicrail XClass locomotive: the maximum sleeper bending stress at the centre is approximately (1 0.45) . 18.5 = 6.5 MPa.
X

 0.36).
(1
(b) For the Westrail LClass locomotive: the maximum sleeper bending stress at the centre is approximately (1 0.36). (1 0.45) . 22.5 = 7.9 MPa.


It is evident that the sleeper dimensions necessary to satisfy the A.R.E.A. design limit for the case of the sleeper bending stress at the centreof the sleeper are far greater than those that are currently used in service by the majority of state railway systems. used Since many state railway systems have also equal to 246 that of the Vicrail XClass axleloadsat least
satisfactorily
for
many
years
in
conjunction
with
track
having
~
sleeper dimensions smaller than those above; it can be concluded that theA.R.E.A. design limit has been defined specifically for the sleeper bending stress at the rail seat. From comparison of Figures 6.10, 6.11 6.13 and 6.14 it is also evident that the Battelle/Schramm method of determining the maximum sleeper bending stress at the rail seat is more conservative than that of the A.R.E.A. method. A.R.E.A. design limit has most conjunction limit The is with unsuitable anomoly for use It is recognised that the probably been derived for method the the and consequently Battelle/Schramm maximum allowable use in
the A.R.E.A. design with that
this
method. sleeper
above
suggests
I
bending stress criterion should be further examined; especially the determinationof reliable estimates of the endurance limit of Australian timber species subjected to the conditions likely to be experienced in service. DEPTH REQUIREMENT
BALLAST
As mentioned in Chapter 5 there are various methods, that are available for use in order to determine the required ballast
depth, viz: theoretical, semiempirical and empirical. In order to simplify the parametric analysis the Schramm method was adopted (Equation 5.15). Using this method the variation of the required ballast depth with the allowable subgrade pressure and with various sleeper spacings for the Vicrail XClass and Westrail is presented in Figures LClass locomotives was determined and 6.16 and 6.17 respectively. This variation was determined for a 2.440 m long and 0.230 m wide sleeper only and obviously if the sleeper bearing dimensions are increased the required ballast depth would be reduced. Referring to Table 5.1 it can be seen that Clarke suggests that the safe average bearing pressure of compacted subgrade is at least 280 kPa. Applying Clarke's suggested factorof safety of 1.67 the ninimum allowable subgrade pressure of compacted subgrades is determined as 168 kPa.
247
The effect of varying the allowable subgrade pressure upon the required ballast depth determined by the Schramm method and a for sleeper spacing of 600 mm is presented in Table 6.14. Of particulal importance is the effect of compaction of the subgrade upon the ballast depth. It can be seen that increasing the allowable subgrade pressure by100 per cent, i.e. from 150 to 300 kPa reduces the required ballast depth by 50 per cent.
The effect of varying the sleeper length 2.440 from to 2.640 m upon the required ballast depth €or particular allowable subgrade pressures can be seen by comparing Figure 6.14 and 6.18. The results of this comparison for a sleeper spacing of 600 mm is presented in Table6.15. It is evident that an 8 per cent increase in the sleeper length results a in reduction of the required ballast depth in the order 16 of per cent. It would seem that increasing the subgrade bearing capacity by means compaction hasa greater benefitin reducing the ballast depth required than by increasing the sleeper size. of
248
TABLE 6.14
 EFFECT
DEPTH
OF VARYING THE ALLOWABLE SUBGRADE PRESSURE UPON THE REQUIRED BALLAST USINGTHE SCHRAMM METHOD
"~
Ballast depth by the Schramm Change in Ballast Depth Allowable Change in subgrade allowable method sleeper for spacing of (Per cent) 600 mm pressure subgrade (mm) ( " a ) pressure (Per cent) VicRail Westrail Due an to increase Due to a XClass Lc lass in allowable theincrease in subgrade pressure axleload
" "
+50
150 225
3 00
+l00
320
370
33
+l6
2 10 160
250 19 0
50
+l9 +l9
Note: Sleeper length
2.440 m, breadth 0.230 m.
TABLE 6.15
 EFFECT
OF VARYING THE SLEEPER LENGTH AND THE ALLOWABLE upo~ THE REQUIRED BALLAST DEPTH USING THE S C H R A " METHOD AT A
SUBGRADE BEARING 600mm SLEEPER
"
PRESSURE
SPACING
_ "
Allowable Change in Ballast depth for VicRail Increase in sleeper Approximate reduction subgrade allowable XClass locomotive (mm) length ballast indepth pressure subgrade pressure (kPa) 2.440 m cent) (Per cent) 2.640 (Per m (Per cent) sleeper sleeper 1length eng th 150 225
300
+5 0 +l00
320 210 160
260 180 130
+8 +8 +8
16
16
16
Note: Sleeper breadth 0.230 m.
m
m
200.
I Clarkes m i n i m u m "li r m i o tf
/ compacted subgrades (168)
100r 100
I
I
125
150
175 200 225 250 Allowable Subgrade Pressure (KPa)
275
Note: Design Conditions Standard gauge track Vicrail Xclass locomotives Speed 100 k m i h A.R.E.A. Impact Factor A.R.E.A.Rail Seat Load Ballast friction angle 35" Sleeper length: 2.440 m Sleeper breadth: 0.230 m
Figure 6 . 16 Variation of the required ballastdepth with allowable subgrade pressure for a range of sleeper spacings, using Schramm's formula, Vicrail Xclass locomotives, and 2.440m long sleepers
251
m
m
300
150125100
300 275 175 250 225 200 Allowable Subgrade Pressure (KPa)
Note: Design Conditions Standard gauge track loco. Westrait Lclass Speed 100 k m / h A.R.E.A. Impact Factor A.R.E.A. Rail Seat Load Ballast friction angle 35" Sleeper length 2.440m Sleeper breadth 0.230m
Figure 6 . 17 Variation of the required ballast depth w i t h allowable subgrade pressure for a range of sleeper spacings, using Schramm's formula, Westrail Lclass locomotives, and 2.440m long sleepers
252
6oo
5
Clarkes m i n i m u m l i m i t for compacted subgrades
100
I
I
,
300
100 175 150 125
275250225200 Allowable Subgrade Pressure (KPa)
Note: Design Conditions Standard gauge track Vicrail Xclass loco. Speed 100 ( k m h ) A.R.E.A. Impact Factor A.R.E.A. Rail Seat Load Ballast friction angle 35" Sleeper length: 2.640 m Sleeper breadth: 0.230 m
Figure 6 . 18 Variation of the required ballast depth w i t h allowable subgrade pressure for a range of sleeper spacings using Schramm's formula, Vicrail Xclass locomotives and 2.640m long sleepers
25 3
CHAPTER 7
 CONCLUSIONS
The current practice for the design of conventional railway trac is entirely based upon satisfying several criteria for the strength of individual components, (Figure 1.1). Due to variations in the track support capacity, which is dependent upon the track maintenance, the design loadings determined by the theoretical and empirical relationships are arbitrarily increased by what are considered reasonable safety factors. Having determined the maximum stresses in the individual components of the track
structure, they are then compared with the corresponding maximu allowable stresses, and if satisfied, the design is completed.
of track maintenance It is clearly apparent that the standard governs the force and stress levels that are applied to the t structure under service conditions.At present the standard of track maintenance is implied in the current track design
procedure by the use of various factors of safety. These factors of safety appear to be based entirely upon the considered judg ment of earlier railway researchers and are always used to predict the maximum force or stress levels likely to occur for the considered worse maintenance condition of the track. The resulting track design is obviously safe although somewhat conservative since the design approach is not related to track maintenance procedures or to the track maintenance condition. Reliable inputs of actual intrack conditions which relate to maintenance procedures are required in the design procedure, or alternatively standards of track maintenance need to be specified in the design method.
In order to introduce a proper engineering approach to railroad track design, consideration of the deformation of the track un traffic loadings is also necessary. Ballast and subgrade properties needto be fully examined and allowable limits of &€ormation based upon the retention of track geometry and maintenance standards need to be evolved. When these factors are included in the conventional design flowchart an interaction chart signifying 254
the design parameter interactions can be developed, and is presof this approach is ented in Figure 7.1. The major advantage that it provides a basis for maintenance and design to be treated as related items since the time dependent (or cyclic load) behaviour of the track support condition is included.
Nevertheless, the existing design expressions permit a degree of optimisation of the track crosssection. The data contained herein can be used for that purpose and an Interactive Track Design Model established.
255
WheelRailInteraction TimeDependent
“ “ “ “ “ c
Rail M o m e n t and
Sleeper Ballast Contact Pressure
Pressure Depth
1
Total Settlement
Ballast and
Figure 7 . 1 Preliminary track design interaction diagram
256
ANNEX A THE THEORETICAL DETERMINATION OF THE TRACK SPRING RATE The track modulus value used in the on beam an elastic foundation analysis and presented in Tables 3.4, 3.5 and 3.6 has been determined for the field condition, i.e., only afterthe track has been constructed. This modulus value lumps the effects of a great numberof parameters together and therefore can only be considered as an average value for the design situation. In order to predict values of the track modulus before the track is constructed a ballast pyramid model has been developed at Battelle Colombus Laboratories [Prause et al. 1974). This model makes use of realistic track structure parameters in its analysis. Talbot (19181934) has shown that the effective ballast friction angle, (i.e. the measure of the degree of load spread of the ballast), changes considerablyat a depth of about 150 mm below the sleeper (Figure A.11. Therefore the ballast layer has been divided into two sections, each having a different friction anglem order to effectively model this observed condition [Figure A.2).
Schramm (1961)
also stated that the load spread angle of ballast
is also dependent upon the (.i.e. the ballast friction angle) condition of the ballast. For ballast in an excellent and poor condition the ballast friction angle is about 4 0 ' and 30° res
pectively, with average values of 35 commonly occurring in 2 m m) track. The expressions which define the effective area ( A(z) at ballast depths of z, for the two sections in Figure A.2 are: (a) Upper Section:
0
A(z] = Cclz + B) (c1z + L) for depths 0 <, z 5 z1 .
24722/8036
257
750
600
450
300
150
0
150
300
450
600
750
Distance from Centreline of Sleeper ( m m )
Figure A . l Lines o f equal vertical pressure (kPa) i n the ballast for a single loaded sleeper (Talbot 1920)
258
Figure A 2 The ballast pyramid
model
259
(b) Lower Section:
where
cl c2 B L
= =
2 tane 2 tane 2,
=
= = 81 = (1)
Z1 =
sleeper breadth (mm), effective sleeper length under the rail seat (mm), angle of internal friction, upper section (degrees),
z2 Zt
2'
= =
angle of internal friction, lower section (degrees), depth of upper section of ballast (mm), depth of lower section of ballast (mm), total depth of ballast layer (mm), and z zl.
to

According
the
Ballast
pyramid
model as developed
by Prause and
of the Ballast Meacham (1974) the stiffnessof the upper section
D1 (.kN/mm) is
and is
the
stiffness
of
the
lower
section
of
the D 2 ballast (kN/mm)
where
Eb
=
modulus of elasticity of the ballast (MPa); typical values ofEb are of the order of 200(MPa).
(1) z1
=
150 (mm).
260
Conventional
Track Structure
Spring
Rates of Track
Continuous Rail
Components
\
“ U +
Beam
on
Continuous Elastic Support
$ 6 9 k s i
\
\
\ \
\
D,
\
Dbl
Figure A 3 Progressive steps i n the development o f a static model o f a conventional track structure (Robnett et al 1975)
261
The stiffnessof the total ballast pyramid Db(kN/mm) can be assumed as the series equivalent of effective stiffnesses of two springs, D1 and D2, in series, i.e.,
Db

D1D2 D1 D2
+
*
The
assumed
mechanical
model
of
the
pad,
sleeper,
ballast
and
A.3. The ballast subgrade is illustrated in Figure spring system Dbs(kN/mm) is denoted by
 subgrade
" 1 Dbs where Dmod
1
Db
+  1
Drnod
I
= = =
the modified soil stiffness (kN/mm),
Dmod
A(z)
'
A
(2)
=
the coefficient of subgrade reaction of the soil 3 (kN/mm , and effective area of the lower section of the balla 2) Equation A. 2. (m
Therefore,
The spring constant of each sleeper Ds(kN/mm) is the series
equivalent of the spring constant of the resilient pad (below the rail seat] D (kN/mm), plus half (due to continuity of the ballast P and the subgrade) of the spring constant of the effective ball
 subgrade
DS
foundationDbs beneath the sleeper, i.e.,
"1 
DP
1 Dbs/2
262
Actual measured total track depressions and the breakdown of the amount of depression caused by the ballast and the subgrade for various subgrade types and due an to axle load of 190 (kN) are presented in Figure A.4 (Birman 1968). It is clearly apparent that the overall track depression and therefore the spring rate is principally governed by the subgrade type. As the subgrade becomes stiffer the ballast contribution to the total depression becomes more significant. The overall foundation modulus k(MPa) is then defined as
k
=
DS 
S
'
(A.10)
'
where
S
=
sleeper spacing (mm). model, the maximum rail
From the beam on an elastic foundation seat load qr(kN) is defined as
(A.11)
where
y ,
=
maximum rail depression (mm).
Combining
Equations A. 6 and A. 7
Therefore the effective the ballastpa(kPa) is
contact
pressure
between
the
sleeper and
P ,
where
B L
=
=
Ds Ym
~
B.L
'
(A.13)
sleeperbreadth (mm), and effective length of the sleeper under therail seat (mm).
263
From this analysis the average pressure uza(kPa) on the subgrade
is defined as
(A.14)
where
A (z)
=
effective area of the lower section of the ballast 2 (m ), Equation A. 2.
264
90
   Rock Gravel Clay Swamp
Subgrade T y p e
5
90
5
W
5
90
5
T r a i n Speed (krnih!
4t
Legend:
0 Elastic Pad
m
Ballast Subgrade Down to 1 (m) Subgrade Below 1 (m)
Figure A4 Track depression for different subgrade types for an axle loadof 190 (kN)(Birrnan 1968)
24722/8037
265
ANNEX B THE ESTIMATION OF THE COEFFICIENT OF SUBGRADE WACTION
Terzaghi (1955) introduced the concept of the "bulb of pressure'' to enable the coefficient of subgrade reaction determined from insitu plate dimensions. test
to be factored according to the actual foundation
The coefficient of subgrade reaction for a soil C,(kN/mm 3 ) is defined as the ratio c, P Y '

where
p
=
pressure per unit of the surface of contact between a loaded beam or slab and the subqrade on which it rests and on which it transfers the load (kPa), and settlement produced by this load application (mm).
y
=
As a basis for estimating the coefficient of subgrade reaction for a loaded area of the subgrade, the coefficient of subgrade 3 reaction C0 (kN/mm ) for a square plate with a width of 1 (ft), 300 (.mm),has been selected, because this value can readily be calculated in the field if necessary.
Terzaghi (1955) suggests that if the subqrade consists of cohesionless or slightly cohesive sand, the value CO of can be estimated from TableB.l. The densitycategory of the sand can be ascertained by means ofa standard penetration test.
These values are valid for contact pressures which are smalle than one half the estimated bearing capacity of the clay; the latter being independent of the dimensions of the loaded area. For a subgrade of heavily precompressed clay the magnitude of CO increases in simple proportion to the unconfined compressive strength of the clay qu(kPa). Terzaghi (1955) has proposed
266
numerical values of CO for various clay types and these are presented TABLE B. 1 in Table B.2. OF C O (kN/mm3 ) FOR SQUARE PLATES 1 FT X 1 FT,
 VALUES
(300 X 300 mm) RESTING ON SAKD (TERZAGHI 1955)
Relative Loose Medium Sand Density Dense of Dry or moist sand, limiting values 21105 721 Dry or moist sand, proposed values 14 8 Submerged sand, proposed values 105350 175 105
45 28
TABLE B.2
 VALUES
(300
X
OF CO 106(kN/mm2) FOR SQUARE PLATES 1 FT X 1 FT 3 0 0 m ) RESTING ON PRECOlWRESSED CLAY (TERZAGHI
1955)
Hard Stiff Consistency Very Stiff Clay of Values of qu(kPa) Range ofCO Proposed valuesCO 107214 17 34 25.5 214428 34 68 51 >428
>
68 102
For sands, experimental investigations have shown that the coefficient of subgrade reaction CS for the effective stressed area of the subgrade , A (z), Equation 5.21, can be related to the measured coefficientof subgrade reaction of the plate test CO using the concept of the "bulb of pressure" by W' + 1 2 ___ 2w '
CS
=
CO
.
267
therefore,
W'
=
c2z' + clzl + B
.
For clayslif the stressed area is of rectangular shape, as is the case with the effective subgrade stressed area, the coeffici C S for this area can be determined by of subgrade reaction
where
X
=
ratio of the width to the length of the effecitve subgrade stressed area, and
The main problem with this method as applied to railway conditi is that the determination of the coefficient of subgrade reacti C O from the field plate bearing test, is based on rather arbi trary static loading conditions (.Prauseet a1 1974). Designs should be based on dynamic properties measured under loading conditions which better approximate the subgrade loading under the track. ,
26 8
BIBLIOGRAPHY Agarwal, M.M. Saharanpur. (1974) Indian Railway Trcick, Manglik Prakashan,
Ahlf, R.E. (1975) Maintenance OF i?czy Ccats, Bow They are A2fected Railway Track and Structures, by Car Weights and Track Structure, Vol. 71 No. 3, p. 34. Amans, F., Sauvage, R. (1969) ?aii,wag Irzc?% SkabilitLd in 8eLaTion
to Transverse Stresses Exertedby RolZ<rcg Stccic
A
Tkeoretical
Study of Track Behaviour, Bulletin of the Intern. Railway Congress Association, 685716.
AREA (1951) AREA Proceedings, Vol. 51, 566. AREA (1952) Effect of FZat WheeZs an Track and Z ? L L < ? ? ~ KAREA ~, Proceedings, Vol 53, 423449. AREA (1973), ManuaZ of Becovmended ?rQc;<ee, American Engineering Association, Illinois. Railway
I
AREA (1975) Manual Recommendatisna, S?scia;! Committee on Concrete Ties, P a r t 10 Concrete Ties (and Fzatenings), AREA Bulletin 655. Bartlett, D.L., (1960) The Stability 'c L.ong WeZded RaiZs, Civil Eng. and Public Works Review, Vol. 55, No. 649, 10331035, NO. 650, 11701171, No. 651, 12991303, No. 653, 15911593. Bhatia, G.S., Romualdi, J.P., Thiers, G.R., (1968) SCxdy of New
Track Designs, U.S. Departnent
0ffransportation,
OffZce o$ gigh
Speed Ground Transportation StQciy. of iiew Track Structure Designs (Contract 22266(reg)), CarnegieMellon University, Transportation Research Institute, Pittsburg, Pennsylvania, U.S.A.
269
Birmann, F. (196566) Track Parameters, Static and Dynamic.
Interaction Between Vehicle and Track, Proc. Inst. Mech. Eng., Vol. 180, Part 3F, 7385.
Birmann, F. (1966) Reihennessungen an der Schiene zur Ermittlung
der Fuhrungskrafte, (Series Measurements at the Rail for Determination of the Guide Forces), in German, ETR Eisenbahntechnische
Rundschau, Vol 15, No. 5, 155164. Birmann, F. (1968)IRCAUIC High Speeds Symposium, Vienna, 1968, Bulletin of the Intern. Railway Congress Association. Boussinesq, J. (1885)Application des Potentiels a Z'Etude de GautheirZ'EquiZibre e t du Mouvement des Solides Elastiques. Villars. Paris, France. (original not seen) Broken Hill Proprietary CO Ltd (1975) BHP Rail Track Material, BHP , Melbourne. Brown, J.H., Skinner, D.H. (1978) The Development o f a Steel Haul Railway Institutionof Conference, Paper X.3,
Sleeper System for Heavy Haul Railways, The
Engineers Australia, Heavy Perth, September 1822.
Clarke, C.W. (1957) Track Loading Fundamentals, The Railway Gazette, Part 1, 4548, Part 2, 103107,Part 3,157163, Part 4, 220221, Part 5, 274278, Part 6, 335336, Part 7, 479481. Clark, P.J. (1973) A n Australian Diesel Locomotive Pocketbook, Australian Railway Historical Society, Sydney. Danzig, J.C., Hay, W.W., Reinschmidt, A.J. (1976) Procedures for
Analysing the Economic Costso f Railroad Roadway for Pricing Purposes, Volumes l and 2, TOPS Online Services Inc.,Report No. RPD11CMR, ContractNo. DOTFR30028.
270
Department of Scientific and Industrial Research, Road Research Laboratory (1961)Soil Mechanics for Road Engineers, Her lrlajesty's Stationery Office,London. Duckworth, J. (1973) Rail Sleepers in Tasmania, 1st International Rail Sleeper Conference, Session 3, Paper 3, Sydney, August1317. Eisenmann, J. , (1969a) Stress on the Permanent Vay dze to fligh
AxZe Loads, Stahle Eisen,Vol. 89, No. 7 , 373.
Eisenmann, J., (1969b) Stresses in the ? a ' % Acting as a Beam, Special excerpt from ETR Eisenbahntechnische Rundshau, Vol. 8, Hestra, Darmstadt. Eisenmann, J., (1969~) Bearspruchu~g ?er Schiene als Prager' (The RaiZ as a Support Being Subject to Stress) in German, ETR Eisenbahntechnische Rundschau,Vol. 8, 306312, Hestra, Darmstadt. Eisenmann, J., (1970) Stress Cistriakzla? in tke ?err?aner,$ ; ' a y
due to Beavy AxZe Loads andSigh Speeds, AREA Proceedings, Vol
,
71, 2459. Eisenmann, J. , (1970) Einflatis daza Sc%zeZIengrosse, ScimeZZeni u f 2 i e Untergrundbeansp~~~chu I, ng teiZting and Schotterbettstarke c
IInfZuence of Sleeper Size and Spacing zxd tiaZ7,ast Depth on the SubsoiZ) in German. ETR Eisenbahntechnische Rundschau, Vol 19, pt 8, 307317, Hestra, Darmstadt.

Engesser, F. , (1888) Zur 3e:pechr,unc des Zisenbahnobepbaues, (To
the AnaZysis of Railroad Track) in German, Organ fur die schritte des Eisenbahnwesens,1888. (original not seen)
Fort
271
Gordon, E.R. (1973) Sleeper Usage in Australia and mew Zealand: Past, Present and Future, 1st International Rail Sleeper Conference, Session 1, Paper 1, Sydney, August1317. Hay, W.W.
N.Y.
(1953) Railroad Engineering, Vol 1, John Wiley & Sons,
Hay, W.W., Schuch, P.M., Frank, M.W., Milskelsen, M.J.
(1973)
Evaluation of Rail Sections, 2nd Progress Report,Civil Eng. No. 442220Transportation SeriesNo. 9, Univ. Illinois Project
332, Univ. Illinois, Urbana. Hay, W.W., Reinschmidt, A.J., Bakes, P.T., Schuch, P.M. (1975)
~
Economic Evaluation of Special Metallurgy Rails, Final Report. Civil Eng. Transportation Series No. 9, Univ. Illinois Project No. 442220332, Univ. Illinois, Urbana.
Heath, D.L., Cottram, M. (1966) Stresses Beneath a Railway The Gazette, Track, Railway
Vol. 124, 10011006.
(1972)
~
Heath, D.L., Shenton, M.J., Sparrow, R.W., Waters, J.M.
Design of Conventional RaiZ Track Foundations,Proc. Inst. of Civil Engineers (London), Vol. 51, 251267.
Hetenyi,'M. (1946) Beams o n Elastic Foundation, The University of Michigan Press., Ann Arbor.
Hoare, J.F., Payne, A.F. (1978) Track Structure for the Kwinana to Koolyanobbing Railway, The Institution of Engineers, Australia, Heavy Haul Railway Conference, Session 403, Paper 5.5, Perth, September.1822. Jenkins, H.H., Stephenson, J.E., Clayton, G.A., Morland, G.W., Lyon, D. (1974) The Effect of track and VehicZe Parameters o n WheeZ/RaiZ Vertical Dynamic Forces, The Railway Engineering Journal, Jan 216.
272
Johnsson, K.L., Jeffries, J.A. (1963) PZas$ic Flow and ResiduaZ Stresses in Rolling and Sliding Contact, Fatigue in Rolling Contact, Session2, Paper 5, Proceedings I. Mech. E., Symposium on Fatigue,5465. Kaiser, W.D., Nessler, G.L., Meacham, H.C., Prause, R.H. (1975) A Review of Measurement Pechniqzes, Beqwirernents and Ava?:labZe Data on the Dynamic Compliaizce of 3 e i c ~ c ~ ~ ~Tnock; ced Inte~irn Report, Battelle Columbus Laboratories, Report No. FRAOR&D7670, National Technical Information Service, Springfield, Virginia, USA. Kerr, A.D. (1976) On the Stress Anallusis of Rails and Ties, AREA Proceedings,Vol. 78, 1943. King, F.E. (1976) Rail Replacement "osts on B.C. South Line, Technical Research, Dept. of Research and Development, Canadian National Railways, Montreal, Canada. Koffman, J.L. (1972) Locomotive Axleloads, Rail Engineering International, Vol 2, 414415. Koffman, J.L., Fairweather, D.M.S. (1975) Some Problems OF Railway Operation at High Axleloads, Rail Engineering International, Vol. 5, 156160. Kurzweil, L. (1972) Dynamic Track CornpZiam?e, U.S. Department of Transportation, Transport Systems Centre, Cambridge Ma., Report No. GSP067, Cambridge, Massachusetts,USA. Lombard, P.C. (1974) Spoorbaanstruktuur: V e r b a d T u s s e n Struktuurparameters Verkeerslading e n Anderhoud, Track Structure: A Relationship Between Structure Parameters, Traffic Loading and Maintenance), Draftof Masters South Africa) in Afrikaans. Thesis, University of Pretoria,
24722/8038
273
Luber, H. (1961) Bin Beitrag Z U T Berechnung des eZastisch getag(A contribution erten Eisenbahnoberbaues 'bei vertikaZer Belastung, to the analysis of the elastically supported railroad track subjected to vertical loads) in German. Dr. Ing. Dissertation, Technical University of Munich. (original not seen). Lundgren, J.R., Martin, G.C., Hay, W.W. (1970) A SimuZation Model of BalZast Support and the ModuZus of Track Elasticity, (Masters Thesis), Civil Engineering Studies, Transportation Series No. 4, Univ. of Illinois. Magee, G.M. (1965) Calculation of Rail Bending Stress for 125Ton Tank Cars, AAR Research Centre, Chicago, Illinois. Magee, G.M. (1965a) Welded Rail on Bridges, Railway Track and Structures, 2 4 2 6. Magee, G.M. (196533), Cooperative Research o n Wood Ties of the
Railroad Tie Association and the Association of American Railroads Research and Test Department,First Progress Report, AAR.
Magee, G.M. (1966), Cooperative Research o n Wood Ties of the
RaiZraod Tie Association and the Association of American Railroads Research and Test Department,Second Progress Report, AAR.
Magee, G.M.
(1967), Cooperative Research o n Wood Ties of the
Railroad Tie Association and the Association of American RaiZroads Research and Test Department,Third Progress Report, AAR.
Magee, G.M. (1968), Cooperative Research on Wood Ties of the of American RaiZroads Railroad Tie Association and the Association
Research and Test Department,Fourth
Progress Report, AAR.
Magee, G.M. (1971), Cooperative Research o n Wood Ties of the and the Association of American RaiZroads RaiZroad Tie Association Research and Test Department,Fifth Progress Report, AAR.
274
Mair, R.I. (1974) Crane Rail Selection, BHP Melb. Res. Labs. Rep. MRL 17/31. Mair, R.I., Groenhout, R. (1975) Ballanced Loading of Ra<Ts (BHPMNM/RDC/75/016). Mair, R.I., Groenhout, R. (1978) Prediction Rail Steel Strength A reZiabiZity Anczlysis, ASTM Symposium, Rail Steels Requirements Developments, Processing and Use, ASTM Special Technical
{ o f

k
Probabilistic Approach, BHP Melb. Res. Labs. Rep. MRL/081/75/010,


Publication No. 644. Mair, R.I., Jupp, R.A., Groenhout, R. (1978) The Characteristics and Control of Longpitch Rail Corrzgation at Heavy Axle Loads, The Institution of Engineers, Australia, Heavy Haul Railway Conference, Session 417, Paper 1.8, Perth, September 1822. Meacham, H.C., Ahlbeck, D.R. (1969) L Computer Study 0; Dynamic Loads Caused by VehicleTrack In$ergction, Trans. ASME, Journal of Eng. for Industry, 69RR1. Okabe, Z. (1961) Laboratory Investigation of Railroad Ballast, Japan Railway Civil Eng. Association, Permanent Way 13. No. Olson, E., Johnsson, S. (1960) Lateral Forces Between Wheels and RaiZs, ASME Paper No. 60RR6.
ORE (1965) Stress'es in Rails, Question 017, Stresses in the Rails,
the Ballast and the Formation Resulting from Traffic Loads, Report D71/RPI/E , Utrecht.
ORE (1966) Stresses i n Rails, Question 017, Stresses i~ the Bails,
the Ballast and the Formation Resulting from TraffSle Loads, Report D71/RP2/EI Utrecht.
ORE (1968) Stresses in tke Formation, , ' ; ~ s s t i o D ~L / , S t r e s s e s in
"
the Rails, the BalZast ar;d the
~ O Z V V C 2ssw5,iq ~ ~ G ~
From I"raf,"ic
Loads, Report D71/RP8/E, Utrecht.
275
ORE (1969) Stresses in Concrete Sleepers, Question 017, Stresses
i n the Rails, the Ballast and the Formation ResuZting from traffic Loads, Report D71/RP9/E, Utrecht.
ORE (1970) Summary Report, Question 071, Stresses in the Rails,
the Ballast and the Formation Resultingfrom Traffic Loads, Report D71/RP13/EI Utrecht.
O'Rourke, M.D. '(1978) Critique of Conventional Track Design BHP Melb. Procedures as Applied to Heavy Axle Load Conditions, Res. Labs Rep. No MRL/C77/78/27(1), (BHP MNM HI/TDC/78/053). Prause, R.H., Meacham, H.C. (1974) Assessment of Design Toozs
and Criteria for Urban RaiZ Track Structures, Vol. l, AiGrade
No.. TieBallast Track, Battelle Columbus Laboratories, UMTA Report UMTAMA060025744, National Technical Information Service, US Dept. of Commerce, Springfield, Virginia. Railway Gazette, (1970) Dynamics o f Railway Track Systems and
Their Economic Consequences, Vol. 126, No. 1, 1924.'
Railway Track and Structures (1970) AAR now has Recommended Track
Standards, Vol. 66, NO. 11, 2831.
Raymond, G.P. (1977) Railroad Wood Tie Design and Behaviour, Proc., ASCE, Trans. Eng. Journal, V103, TE4, 521. Read, R.G. (1972) The Rail for High Intensity MineraZ Traffic, AREA Bulletin 639, 3851. Reid, D.A.C. (1973) Test InstaZlations of Treated Softwood
1st International Rail SZeepers in the South Australian Railways, Sleeper Conference, Session 5, Paper 2 , Sydney, August1317.
276
Robnett, Q.L., Thompson, M.R., Hay, W.W., Flayabji, S.D., Peterson, H.C., Knutson, R.M., Baugher, R.W. (1975) Technics% Data 3ases Report, Ballast and Foundation Materials Research Program, FRA Report No. FRAOR&D76138, National Technical Information Service, Springfield, Virginia,USA.
Schramm, G. (1961) Perrnczne~tF l a y i"ezh<sue Verlagsgesellschaft Darmstadt.
a i 7 . d :Fzrmme?:t
Xciy
Economy, (English Translation by Hans Lange), Otto Elsner
Schwedler, J.W. (1882), On Iron Perganent Way, Proc. Inst. of Civil Eng's. London (original not seen). Shenton, M.J., Powell, M.C. (1973) ?>e Resistance Gfpe?.e? by the Ba%last to Lateral Movement 5F S%zz?e~rs, Summary C S 2 . 3 Tests, OR6 D117 (unpublished draft). Sinha, H. (1967) Ballast in Rai'Lwag.Track, Indian Railway Technical Bulletin. 1526. Smith, J.O. (1942) T h e 37ffect of Zan,e 0 . f Stress on the Fatigue Strength of !detals, University of Illinois Engineering Experiment Station Bulletin, Ser. No. 334. Smith, J.O. , Liu, C.K. (1953) Skresaes dxa t~ Tangential axd
Normal Loads o n an Tlastic SoiZd tiith LupSicaztoq to some Cgntact Stress Problems, Trans. ASME, J. App. Mech, 157166.
Srinivasan, M. (1969) I,!Dderz rermanent %'ay, Somaiya Publications Pty Ltd,Bombay.

Talbot, A.N. (Chairman) et all (1918) Stresses < X L7ailrocd T'rack, Report of the SpeciaL Cammitfae t c Report on Stresses in Rai%road Track; First Progress Report', AREA Proceedings, Vol. 19, 8731062. 277
Talbot, A.N.
(Chairman) et al, (1919) Stresses in RaiZroad Track,
Report of the Special Committee to Reporto n Stresses i n Railroad Track; Second Progress Report', AREA Proceedings, Vol. 21, 645814.
Talbot, A.N. (Chairman) et al, (1922) Stresses in Railroad Track, Report of the Special Committee to Report o n Stresses in Railroad Vol. 24, 297453. Track; Third Progress Report', AREA Proceedings, Talbot, A.N. (Chairman) et al, (1925) Stresses in RaiZroad Track, o n Stresses in Railroad Report of the Special Committee to Report Vol. 26, 1081Track; Fourth Progress Report', AREA Proceedings,
1245.
Talbot, A.N.
(Chairman) et al, (1929) Stresses in Railroad Track,
Report of the Special Committee to Report o n Stresses i n Railroad Track; Fifth Progress Report', AREA Proceedings, Vol. 36, 69336.
Talbot, A.N.
(Chairman) et al, (1933) Stresses in RaiZroad Track,
Report of the SpeciaZ Committee to Reporto n Stresses i n Railroad Track; Sixth Progress Report', AREA Proceedings, Vol. 45, 66848.
Talbot, A.N.
(Chairman) et al, (1941) Stresses in RaiZroad Track,
Report of the Special Committee to Report o n Stresses i n Railroad Track; Seventh Progress Report', AREA Proceedings, Vol. 42,
753850.
Taute, S., Botha, G.J., Therow, J.G., Hindson, T.R., Viljoen, C.R., McLintock, A.A. (1971) On Axle Loading and Related Matters, Report of the Combined Study Missions of the South African Railways and ISCOR Railways. for the Proposed Sishen Saldanha Railway, South
Afr
Terzaghi, K. (1955) Evaluation of Coefficients of Subgrade Reaction, Goetechnique, 297326.
278
Timoshenko, S. and Langer, B.F. (1932) Siresses in RaiZroad Track, Trans. ASME, Applied Mechanics, Vol. 54, 277293. Vink, P. (1978) New Zealand Looks at Yorsion Resistance, Rail 2, No. 1, 1719. Track, Journal of Rail Sleeper Assoc. (Aust), Vol. Wasiutynski, A. (1937) Reserches ExperixeqtalEs sir Zes Dsformations EZastiques et L e PrauciZ #de Za Sxszrstructure d e s Chemins de fer, Annales de le Academic des Sciences Techniques a Varsovic,
Tome IV. (Original not seen). Waters, J.M., Shenton, M.J. (1968) Trir’c’c Foundation Design, The Railway Gazette,Vol. 124. Weber, J.W. (1975) Development of t h e Trestressed Concrete Tie in
’
the USA, Proceedings of a Symposium held at Princeton University,
April 2123, 1975, Kerr, A.D. (ed) in Railroad Track Mechanics and Technology, Pergamon Press, Oxford, 1978. Westrail, MainZine Upgrading Study 73751376, Interim Report, Standard Gauge Main Line, KwinanaKoolyanobbing Section,
1976. Oct
Winkler, E. (1967) Die Lehre uon de^ Ztasticitat und Testigheit,
(EZasticity and Strength), in German, Verlag von H. Dominicus, Prague (original not seen).
Zimmerman, H. (1888) Die V e r e c h n x z g #&S Zisenbahnaberbaues (The Analysis of the Railroad Track) in German, Verlaq W. Ernst and Sohn, Berlin. (original not seen).
279
ABBREVIATIONS
The following principal abbreviations occur throughout the report AAR : AREA : Batelle: BR : DB : JNR : ORE : SNCF : CWR : Association of American Railroads American Railroad Engineering Association Battelle Columbus Laboratories (USA) British Railways
Deutsche Bundesbahn (German Federal Railways) Japanese National Railways Office for Research and Experiments of the Internatio Union of Railways Societe Nationalede Ehemins National Railways) Continuously Welded Rail. de Fer Francaise (French
24722/80L
280
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue listening from where you left off, or restart the preview.